Researchers from the Department of Mechanical Engineering at Virginia Tech have introduced a continuous fiber reinforcement (CFR) deposition tool designed for multi-axis 3D printing, significantly enhancing mechanical performance in composite structures. Led by Kieran D. Beaumont, Joseph R. Kubalak, and Christopher B. Williams, and published in Springer Nature Link, the study demonstrates an 820% improvement in maximum load capacity compared to conventional planar short carbon fiber (SCF) 3D printing methods. This tool integrates three key functions: reliable fiber cutting and re-feeding, in situ fiber volume fraction control, and a slender collision volume to support complex multi-axis toolpaths. The newly developed deposition tool addresses critical challenges in CFR additive manufacturing. It is capable of cutting and re-feeding continuous fibers during travel movements, a function required to create complex geometries without material tearing or print failure. In situ control of fiber volume fraction is also achieved by adjusting the polymer extrusion rate. A slender geometry minimizes collisions between the tool and the printed part during multi-axis movements. The researchers designed the tool to co-extrude a thermoplastic polymer matrix with a continuous carbon fiber (CCF) towpreg. This approach allowed reliable fiber re-feeding after each cut and enabled printing with variable fiber content within a single part. The tool’s slender collision volume supports increased range of motion for the robotic arm used in the experiments, allowing alignment of fibers with three-dimensional load paths in complex structures. The six Degree-of-Freedom Robotic Arm printing a multi-axis geometry from a CFR polymer composite. Photo via Springer Nature Link. Mechanical Testing Confirms Load-Bearing Improvements Mechanical tests evaluated the impact of continuous fiber reinforcement on polylactic acid (PLA) parts. In tensile tests, samples reinforced with continuous carbon fibers achieved a tensile strength of 190.76 MPa and a tensile modulus of 9.98 GPa in the fiber direction. These values compare to 60.31 MPa and 3.01 GPa for neat PLA, and 56.92 MPa and 4.30 GPa for parts containing short carbon fibers. Additional tests assessed intra-layer and inter-layer performance, revealing that the continuous fiber–reinforced material had reduced mechanical properties in these orientations. Compared to neat PLA, intra-layer tensile strength and modulus dropped by 66% and 63%, respectively, and inter-layer strength and modulus decreased by 86% and 60%. Researchers printed curved tensile bar geometries using three methods to evaluate performance in parts with three-dimensional load paths: planar short carbon fiber–reinforced PLA, multi-axis short fiber–reinforced samples, and multi-axis continuous fiber–reinforced composites. The multi-axis short fiber–reinforced parts showed a 41.6% increase in maximum load compared to their planar counterparts. Meanwhile, multi-axis continuous fiber–reinforced parts absorbed loads 8.2 times higher than the planar short fiber–reinforced specimens. Scanning electron microscopy (SEM) images of fracture surfaces revealed fiber pull-out and limited fiber-matrix bonding, particularly in samples with continuous fibers. Schematic illustration of common continuous fiber reinforcement–material extrusion (CFR-MEX) modalities: in situ impregnation, towpreg extrusion, and co-extrusion with towpreg. Photo via Springer Nature Link. To verify the tool’s fiber cutting and re-feeding capability, the researchers printed a 100 × 150 × 3 mm rectangular plaque that required 426 cutting and re-feeding operations across six layers. The deposition tool achieved a 100% success rate, demonstrating reliable cutting and re-feeding without fiber clogging. This reliability is critical for manufacturing complex structures that require frequent travel movements between deposition paths. In situ fiber volume fraction control was validated through printing a rectangular prism sample with varying polymer feed rates, road widths, and layer heights. The fiber volume fractions achieved in different sections of the part were 6.51%, 8.00%, and 9.86%, as measured by cross-sectional microscopy and image analysis. Although lower than some literature reports, the researchers attributed this to the specific combination of tool geometry, polymer-fiber interaction time, and print speed. The tool uses Anisoprint’s CCF towpreg, a pre-impregnated continuous carbon fiber product with a fiber volume fraction of 57% and a diameter of 0.35 mm. 3DXTECH’s black PLA and SCF-PLA filaments were selected to ensure consistent matrix properties and avoid the influence of pigment variations on mechanical testing. The experiments were conducted using an ABB IRB 4600–40/2.55 robotic arm equipped with a tool changer for switching between the CFR-MEX deposition tool and a standard MEX tool with an elongated nozzle for planar prints. Deposition Tool CAD and Assembly. Photo via Springer Nature Link. Context Within Existing Research and Future Directions Continuous fiber reinforcement in additive manufacturing has previously demonstrated significant improvements in part performance, with some studies reporting tensile strengths of up to 650 MPa for PLA composites reinforced with continuous carbon fibers. However, traditional three-axis printing methods restrict fiber orientation to planar directions, limiting these gains to within the XY-plane. Multi-axis 3D printing approaches have demonstrated improved load-bearing capacity in short-fiber reinforced parts. For example, multi-axis printed samples have shown failure loads several times higher than planar-printed counterparts in pressure cap and curved geometry applications. Virginia Tech’s tool integrates multiple functionalities that previous tools in literature could not achieve simultaneously. It combines a polymer feeder based on a dual drive extruder, a fiber cutter and re-feeder assembly, and a co-extrusion hotend with adjustable interaction time for fiber-polymer bonding. A needle-like geometry and external pneumatic cooling pipes reduce the risk of collision with the printed part during multi-axis reorientation. Measured collision volume angles were 56.2° for the full tool and 41.6° for the hotend assembly. Load-extension performance graphs for curved tensile bars. Photo via Springer Nature Link. Despite these advances, the researchers identified challenges related to weak bonding between the fiber and the polymer matrix. SEM images showed limited impregnation of the polymer into the fiber towpreg, with the fiber-matrix interface remaining a key area for future work. The study highlights that optimizing fiber tow sizing and improving the fiber-polymer interaction time during printing could enhance inter-layer and intra-layer performance. The results also suggest that advanced toolpath planning algorithms could further leverage the tool’s ability to align fiber deposition along three-dimensional load paths, improving mechanical performance in functional parts. The publication in Springer Nature Link documents the full design, validation experiments, and mechanical characterization of the CFR-MEX tool. The work adds to a growing body of research on multi-axis additive manufacturing, particularly in combining continuous fiber reinforcement with complex geometries. Take the 3DPI Reader Survey — shape the future of AM reporting in under 5 minutes. Ready to discover who won the 20243D Printing Industry Awards? Subscribe to the 3D Printing Industry newsletter to stay updated with the latest news and insights. Featured photo shows the six Degree-of-Freedom Robotic Arm printing a multi-axis geometry. Photo via Springer Nature Link. Anyer Tenorio Lara Anyer Tenorio Lara is an emerging tech journalist passionate about uncovering the latest advances in technology and innovation. With a sharp eye for detail and a talent for storytelling, Anyer has quickly made a name for himself in the tech community. Anyer's articles aim to make complex subjects accessible and engaging for a broad audience. In addition to his writing, Anyer enjoys participating in industry events and discussions, eager to learn and share knowledge in the dynamic world of technology.

html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd" The Faculté de l'aménagement at the Université de Montréal is pleased to announce the launch of an international, multidisciplinary and anonymous ideas competition, reserved for students, to create inclusive experiences at the CEPSUM, the Université de Montréal's sports center.With a total of $31,500 in prizes, the competition promotes the idea of invisible accessibility, an experience of the built environment that is of high quality to all, where the design of accessibility is integrated in an indistinguishable manner, and where universal accessibility is envisaged as a global state of the project experience, rather than a dedicated path made up of identifiable and visible solutions.Participants are invited to propose transformative ideas that offer inclusive and equitable experiences for all users. The competition is structured around three typical sports center experiences that are not currently universally accessible:1. The main entrance - Rethinking the entrance and reception of the sports center;2. Carabins stadium - Improve the game-going experience;3. The pool - Creating an inclusive swimming experience.The proposals received over the summer will be evaluated by a multidisciplinary jury of eight experts. For each of the three experiences, three winning projects will be selected, making a total of nine winners.All proposals will be presented in October 2025 at a conference organized by the Faculté de l'aménagement, bringing together researchers working on accessibility in the built environment."We warmly welcome students from around the world to propose bold, creative ideas that reimagine universal accessibility—not as an add-on, but as an integral, seamless, and uplifting experience for everyone, says Carmela Cucuzzella, Dean of the Faculté de l'aménagement. We are looking for designs that are not only inclusive but also invisible in their accommodation, free of stigma, and full of delight and safety. Think beyond the box—then break it wide open.""A public space that is not accessible to everyone cannot be considered public, says Bechara Helal, Associate Dean of Research and Scientific Life. It is high time to rethink the place of universal accessibility in design disciplines, and that is what this competition aims to do: define innovative ways of designing the built environment so that it can become the setting for quality public experiences for all."About the Faculty of Environmental DesignThe mission of the Faculty of Environmental Design is to train high-calibre professionals and researchers who are qualified to contribute to progress and innovation in design practices. The Faculty is recognized for offering students a rich and stimulating learning environment in a context of intellectual freedom in which students acquire innovative work methods and develop the kind of critical thinking and professional discipline that allow them to become responsible citizens aware of the issues facing them.PrizesAn eight-member multidisciplinary jury will review the proposals. There will be three winning projects given out for each of the three experiences:First place (gold): CAD $6,000.Second place (silver): CAD $3,000.Bronze third place: $1,500.00 CADAdditionally, each winner will get a certificate of merit.Registrations are open until July 1, 2025. Submission deadline is August 05, 2025 02:00 PM. The top image in the article courtesy of The Faculté de l'aménagement. > via University of Montreal - Faculty of Environmental Design

As a Canadian who has spent the last two and a half years working as an intern architect in Helsinki, these questions have been on my mind. In my current role, I have had the opportunity to participate in numerous architectural competitions arranged by Finnish municipalities and public institutions. It has been my observation that the Finnish system of open, anonymous architectural competitions consistently produces elegant and highly functional public buildings at reasonable cost and at great benefit to the lives of the everyday people for whom the projects are intended to serve. Could Canada benefit from the adoption of a similar model? ‘Public project’ has never been a clearly defined term and may bring to mind the image of a bustling library for some while conjuring the image of a municipal power substation for others. In the context of this discussion, I will use the term to refer to projects that are explicitly in-service of the broader public such as community centres, museums, and other cultural venues. Finland’s architectural competition system Frequented by nearly 2 million visitors per year, the Oodi Central Library in Helsinki, Finland, has become a thriving cultural hub and an internationally recognized symbol of Finnish design innovation. Designed by ALA Architects, the project was procured through a 2-stage, open, international architectural competition. Photo by Ninara (flickr, CC BY 2.0) In Finland, most notable public projects begin with an architectural competition. Some are limited to invited participants only, but the majority of these competitions are open to international submissions. Importantly, the authors of any given proposal remain anonymous with regards to the jury. This ensures that all proposals are evaluated purely on quality without bias towards established firms over lesser known competitors. The project budget is known in advance to the competition entrants and cost feasibility is an important factor weighed by the jury. However, the cost for the design services to be procured from the winning entry is fixed ahead of time, preventing companies from lowballing offers in the hopes of securing an interesting commission despite the inevitable compromises in quality that come with under-resourced design work. The result: inspired, functional public spaces are the norm, not the exception. Contrasted against the sea of forgettable public architecture to be found in cities large and small across Canada, the Finnish model paints a utopic picture. Several award-winning projects in my current place of employment in Helsinki have been the result of successes in open architectural competitions. The origin of the firm itself stemmed from a winning competition entry for a church in a small village submitted by the firm’s founder while he was still completing his architectural studies.  At that time, many architecture firms in Finland were founded in this manner with the publicity of a competition win serving as a career launching off point for young architects. While less common today, many students and recent graduates still participate in these design competitions. On the occasion that a young practitioner wins a competition, they are required to assemble a team with the necessary expertise and qualifications to satisfy the requirements of the jury. I believe there is a direct link between the high architectural quality outcomes of these competitions and the fact that they are conducted anonymously. The opening of these competitions to submissions from companies outside of Finland further enhances the diversity of entries and fosters international interest in the goings-on of Finland’s architectural scene. Nonetheless, it is worth acknowledging that exemplary projects have also resulted from invited and privately organized competitions. Ultimately, the mindset of the client, the selection of an appropriate jury, and the existence of sufficient incentives for architects to invest significant time in their proposals play a more critical role in shaping the quality of the final outcome. Tikkurila Church and Housing in Vantaa, Finland, hosts a diverse range of functions including a café, community event spaces and student housing. Designed by OOPEAA in collaboration with a local builder, the project was realized as the result of a competition organized by local Finnish and Swedish parishes. Photo by Marc Goodwin Finland’s competition system, administered by the Finnish Association of Architects (SAFA), is not limited to major public projects such as museums, libraries and city halls. A significant number of idea competitions are organized seeking compelling visions for urban masterplans. The quality of this system has received international recognition. To quote a research paper from a Swedish university on the structure, criteria and judgement process of Finnish architectural competitions, “The Finnish (competition) experience can provide a rich information source for scholars and students studying the structure and process of competition system and architectural judgement, as well as those concerned with commissioning and financing of competitions due to innovative solutions found in the realms of urban revitalization, poverty elimination, environmental pollution, cultural and socio-spatial renewals, and democratization of design and planning process.” This has not gone entirely under the radar in Canada. According to the website of the Royal Architectural Institute of Canada (RAIC), “Competitions are common in countries such as Finland, Ireland, the United Kingdom, Australia, and New Zealand. These competitions have resulted in a high quality of design as well as creating public interest in the role of architecture in national and community life.” Canada’s architectural competition system In Canada, the RAIC sets general competition guidelines while provincial and territorial architect associations are typically responsible for the oversight of any endorsed architectural competition. Although the idea of implementing European architectural competition models has been gaining traction in recent years, competitions remain relatively rare even for significant public projects. While Canada is yet to fully embrace competition systems as a powerful tool for ensuring higher quality public spaces, success stories from various corners of the country have opened up constructive conversations. In Edmonton, unconventional, competitive procurement efforts spearheaded by city architect Carol Belanger have produced some remarkable public buildings. This has not gone unnoticed in other parts of the country where consistent banality is the norm for public projects. Jasper Place Branch Library designed by HCMA and Dub Architects is one of several striking projects in Edmonton built under reimagined commissioning processes which broaden the pool of design practices eligible to participate and give greater weight to design quality as an evaluation criterion. Photo by Hubert Kang The wider applicability of competition systems as a positive mechanism for securing better public architecture has also started to receive broader discussion. In my hometown of Ottawa, this system has been used to procure several powerful monuments and, more recently, to select a design for the redevelopment of a key city block across from Parliament Hill. The volume and quality of entries, including from internationally renowned architectural practices, attests to the strengths of the open competition format. Render of the winning entry for the Block 2 Redevelopment in Ottawa. This 2-stage competition was overseen directly by the RAIC. Design and render by Zeidler Architecture Inc. in cooperation with David Chipperfield Architects. Despite these successes, there is significant room for improvement. A key barrier to wider adoption of competition practices according to the RAIC is “…that potential sponsors are not familiar with competitions or may consider the competition process to be complicated, expensive, and time consuming.” This is understandable for private actors, but an unsatisfactory answer in the case of public, tax-payer funded projects. Finland’s success has come through the normalization of competitions for public project procurement. We should endeavour to do the same. Maintaining design contribution anonymity prior to jury decision has thus far been the exception, not the norm in Canada. This reduces the credibility of the jury without improving the result. Additionally, the financing of such competitions has been piece-meal and inconsistent. For example, several world-class schools have been realized in Quebec as the result of competitions funded by a provincial investment.  With the depletion of that fund, it is no longer clear if any further schools will be commissioned in Quebec under a similar model. While high quality documentation has been produced, there is a risk that developed expertise will be lost if the team of professionals responsible for overseeing the process is not retained. École du Zénith in Shefford, Quebec, designed by Pelletier de Fontenay + Leclerc Architectes is one of six elegant and functional schools commission by the province through an anonymous competition process. Photo by James Brittain A path forward Now more than ever, it is essential that our public projects instill in us a sense of pride and reflect our uniquely Canadian values. This will continue to be a rare occurrence until more ambitious measures are taken to ensure the consistent realization of beautiful, innovative and functional public spaces that connect us with one another. In service of this objective, Canada should incentivize architectural competitions by normalizing their use for major public projects such as national museums, libraries and cultural centres. A dedicated Competitions Fund could be established to support provinces, territories and cities who demonstrate initiative in the pursuit of more ambitious, inspiring and equitable public projects. A National Competitions Expert could be appointed to ensure retention and dissemination of expertise. Maintaining the anonymity of competition entrants should be established as the norm. At a moment when talk of removing inter-provincial trade barriers has re-entered public discourse, why not consider striking down red tape that prevents out-of-province firms from participating in architectural competitions? Alas, one can dream. Competitions are no silver bullet. However, recent trials within our borders should give us confidence that architectural competitions are a relatively low-risk, high-reward proposition. To this end, Finland’s open, anonymous competition system offers a compelling case study from which we would be well served to take inspiration. Isaac Edmonds is a Canadian working for OOPEAA – Office for Peripheral Architecture in Helsinki, Finland. My observations of the Finnish competition system’s ability to consistently produce functional, beautiful buildings inform my interest in procurement methods that elevate the quality of our shared public realm. The post Op-Ed: Could Canada benefit from adopting Finland’s architectural competition system? appeared first on Canadian Architect.

More enterprises have moved to pay-for-performance salary and promotion models that measure progress toward goals -- but how do you measure goals for a maintenance programmer who barrels through a request backlog but delivers marginal value for the business, or for a business analyst whose success is predicated on forging intangibles like trust and cooperation with users so things can get done? It’s an age-old question facing companies, now that 77% of them use some type of pay-for-performance model. What are some popular pay-for-performance use cases? A factory doing piece work that pays employees based upon the number of items they assemble. A call center that pays agents based on how many calls they complete per day. A bank teller who gets rewarded for how many customers they sign up for credit cards. An IT project team that gets a bonus for completing a major project ahead of schedule. The IT example differs from the others, because it depends on team and not individual execution, but there nevertheless is something tangible to measure. The other use cases are more clearcut -- although they don’t account for pieces in the plant that were poorly assembled in haste to make quota and had to be reworked, or a call center agent who pushes calls off to someone else so they can end their calls in six minutes or less, or the teller who signs up X number of customers for credit cards, although two-thirds of them never use the credit card they signed up for. Related:In short, there are flaws in pay-for-performance models just as there are in other types of compensation models that organizations use. So, what’s the best path for IT for CIOs who want to implement pay for performance? One approach is to measure pay for performance based upon four key elements: hard results, effort, skill, and communications. The mix of these elements will vary, depending on the type of position each IT staff member performs. Here are two examples of pay per performance by position: 1. Computer maintenance programmers and help desk specialists Historically, IT departments have used hard numbers like how many open requests a computer maintenance programmer has closed, or how many calls a help desk employee has solved. There is merit in using hard results, and hard results should be factored into performance reviews for these individuals -- but hard numbers don’t tell the whole story.  For example, how many times has a help desk agent gone the extra mile with a difficult user or software bug, taking the time to see the entire process through until it is thoroughly solved? lf the issue was of a global nature, did the Help Desk agent follow up by letting others who use the application know that a bug was fixed? For the maintenance programmer who has completed the most open requests, which of these requests really solved a major business pain point? For both help desk and maintenance programming employees, were the changes and fixes properly documented and communicated to everyone with a need to know? And did these employees demonstrate the skills needed to solve their issues? Related:It’s difficult to capture hard results on elements like effort, communication and skills, but one way to go about it is to survey user departments on individual levels of service and effectiveness. From there, it’s up to IT managers to determinate the “mix” of hard results, effort, communication and skills on which the employee will be evaluated, and to communicate upfront to the employee what the pay for performance assessment will be based on. 2. Business analysts and trainers Business analysts and trainers are difficult to quantify in pay for performance models because so much of their success depends upon other people. A business analyst can know everything there is to know about a particular business area and its systems, but if the analyst is working with unresponsive users, or lacks the soft skills needed to communicate with users, the pay for performance can’t be based upon the technology skillset alone.  Related:IT trainers face a somewhat different dilemma when it  comes to performance evaluation: they can produce the training that new staff members need before staff is deployed on key projects,  but if a project gets delayed and this causes trainees to lose the knowledge that they learned, there is little the trainer can do aside from offering a refresher course. Can pay for performance be used for positions like these? It’s a mixed answer. Yes, pay per performance can be used for trainers, based upon how many individuals the trainer trains and how many new courses the trainer obtains or develops. These are the hard results. However, since so much of training’s execution depends upon other people downstream, like project managers who must start projects on time so new skills aren’t lost,  managers of training should also consider pay for performance elements such as effort (has the trainer consistently gone the extra mile to make things work?), skills and communication.  In sum, for both business analysts and trainers, there are hard results that can be factored into a pay for performance formula, but there is also a need to survey each position’s “customers” -- those individuals (and their managers) who utilized the business analyst’s or trainer’s skills and products to accomplish their respective objectives in projects and training. Were these user-customers satisfied?  Summary Remarks The value that IT employees contribute to overall IT and to the business at large is a combination of tangible and intangible results. Pay for performance models are well suited to gauge tangible outcomes, but they fall short when it comes to the intangibles that could be just as important. Many years ago, when Pat Riley was coaching the Los Angeles Lakers, an interviewer asked what type of metrics he used when he measured the effectiveness of individual players on the basketball court. Was it the number of points, rebounds, or assists? Riley said he used an “effort" index. For example, how many times did a player go up to get a rebound, even if he didn’t end up with the ball? Riley said the effort individual players exhibited mattered, because even if they didn’t get the rebound, they were creating situations so someone else on the team could. IT is similar. It’s why OKR International, a performance consultancy, stated “Intangibles often create or destroy value quietly -- until their impact is too big to ignore. In the long run, they are the unseen levers that determine whether strategy thrives or withers.”  What CIOs and IT leadership can do when they use pay for performance is to assure that hard results, effort, communications and skills are appropriately blended for each IT staff position, and its responsibilities and realities -- because you can’t attach a numerical measurement to everything -- but you can observe visible changes that begin to manifest when a business analyst turns around what has been a hostile relationship with a user department and you begin to get things done. 

More enterprises have moved to pay-for-performance salary and promotion models that measure progress toward goals -- but how do you measure goals for a maintenance programmer who barrels through a request backlog but delivers marginal value for the business, or for a business analyst whose success is predicated on forging intangibles like trust and cooperation with users so things can get done? It’s an age-old question facing companies, now that 77% of them use some type of pay-for-performance model. What are some popular pay-for-performance use cases? A factory doing piece work that pays employees based upon the number of items they assemble. A call center that pays agents based on how many calls they complete per day. A bank teller who gets rewarded for how many customers they sign up for credit cards. An IT project team that gets a bonus for completing a major project ahead of schedule. The IT example differs from the others, because it depends on team and not individual execution, but there nevertheless is something tangible to measure. The other use cases are more clearcut -- although they don’t account for pieces in the plant that were poorly assembled in haste to make quota and had to be reworked, or a call center agent who pushes calls off to someone else so they can end their calls in six minutes or less, or the teller who signs up X number of customers for credit cards, although two-thirds of them never use the credit card they signed up for. Related:In short, there are flaws in pay-for-performance models just as there are in other types of compensation models that organizations use. So, what’s the best path for IT for CIOs who want to implement pay for performance? One approach is to measure pay for performance based upon four key elements: hard results, effort, skill, and communications. The mix of these elements will vary, depending on the type of position each IT staff member performs. Here are two examples of pay per performance by position: 1. Computer maintenance programmers and help desk specialists Historically, IT departments have used hard numbers like how many open requests a computer maintenance programmer has closed, or how many calls a help desk employee has solved. There is merit in using hard results, and hard results should be factored into performance reviews for these individuals -- but hard numbers don’t tell the whole story.  For example, how many times has a help desk agent gone the extra mile with a difficult user or software bug, taking the time to see the entire process through until it is thoroughly solved? lf the issue was of a global nature, did the Help Desk agent follow up by letting others who use the application know that a bug was fixed? For the maintenance programmer who has completed the most open requests, which of these requests really solved a major business pain point? For both help desk and maintenance programming employees, were the changes and fixes properly documented and communicated to everyone with a need to know? And did these employees demonstrate the skills needed to solve their issues? Related:It’s difficult to capture hard results on elements like effort, communication and skills, but one way to go about it is to survey user departments on individual levels of service and effectiveness. From there, it’s up to IT managers to determinate the “mix” of hard results, effort, communication and skills on which the employee will be evaluated, and to communicate upfront to the employee what the pay for performance assessment will be based on. 2. Business analysts and trainers Business analysts and trainers are difficult to quantify in pay for performance models because so much of their success depends upon other people. A business analyst can know everything there is to know about a particular business area and its systems, but if the analyst is working with unresponsive users, or lacks the soft skills needed to communicate with users, the pay for performance can’t be based upon the technology skillset alone.  Related:IT trainers face a somewhat different dilemma when it  comes to performance evaluation: they can produce the training that new staff members need before staff is deployed on key projects,  but if a project gets delayed and this causes trainees to lose the knowledge that they learned, there is little the trainer can do aside from offering a refresher course. Can pay for performance be used for positions like these? It’s a mixed answer. Yes, pay per performance can be used for trainers, based upon how many individuals the trainer trains and how many new courses the trainer obtains or develops. These are the hard results. However, since so much of training’s execution depends upon other people downstream, like project managers who must start projects on time so new skills aren’t lost,  managers of training should also consider pay for performance elements such as effort (has the trainer consistently gone the extra mile to make things work?), skills and communication.  In sum, for both business analysts and trainers, there are hard results that can be factored into a pay for performance formula, but there is also a need to survey each position’s “customers” -- those individuals (and their managers) who utilized the business analyst’s or trainer’s skills and products to accomplish their respective objectives in projects and training. Were these user-customers satisfied?  Summary Remarks The value that IT employees contribute to overall IT and to the business at large is a combination of tangible and intangible results. Pay for performance models are well suited to gauge tangible outcomes, but they fall short when it comes to the intangibles that could be just as important. Many years ago, when Pat Riley was coaching the Los Angeles Lakers, an interviewer asked what type of metrics he used when he measured the effectiveness of individual players on the basketball court. Was it the number of points, rebounds, or assists? Riley said he used an “effort" index. For example, how many times did a player go up to get a rebound, even if he didn’t end up with the ball? Riley said the effort individual players exhibited mattered, because even if they didn’t get the rebound, they were creating situations so someone else on the team could. IT is similar. It’s why OKR International, a performance consultancy, stated “Intangibles often create or destroy value quietly -- until their impact is too big to ignore. In the long run, they are the unseen levers that determine whether strategy thrives or withers.”  What CIOs and IT leadership can do when they use pay for performance is to assure that hard results, effort, communications and skills are appropriately blended for each IT staff position, and its responsibilities and realities -- because you can’t attach a numerical measurement to everything -- but you can observe visible changes that begin to manifest when a business analyst turns around what has been a hostile relationship with a user department and you begin to get things done. 

June 3, 20258 min readWhite House Budget Plan Would Devastate U.S. Space ScienceScientists are rallying to reverse ruinous proposed cuts to both NASA and the National Science FoundationBy Nadia Drake edited by Lee BillingsFog shrouds the iconic Vehicle Assembly Building at NASA’s Kennedy Space Center in Florida in this photograph from February 25, 2025. Gregg Newton/AFP via GettyLate last week the Trump Administration released its detailed budget request for fiscal year 2026 —a request that, if enacted, would be the equivalent of carpet-bombing the national scientific enterprise.“This is a profound, generational threat to scientific leadership in the United States,” says Casey Dreier, chief of space policy at the Planetary Society, a science advocacy group. “If implemented, it would fundamentally undermine and potentially devastate the most unique capabilities that the U.S. has built up over a half-century.”The Trump administration’s proposal, which still needs to be approved by Congress, is sure to ignite fierce resistance from scientists and senators alike. Among other agencies, the budget deals staggering blows to NASA and the National Science Foundation (NSF), which together fund the majority of U.S. research in astronomy, astrophysics, planetary science, heliophysics and Earth science —all space-related sciences that have typically mustered hearty bipartisan support.On supporting science journalismIf you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.The NSF supports ground-based astronomy, including such facilities as the Nobel Prize–winning gravitational-wave detectors of the Laser Interferometer Gravitational-Wave Observatory (LIGO), globe-spanning arrays of radio telescopes, and cutting-edge observatories that stretch from Hawaii to the South Pole. The agency faces a lethal 57 percent reduction to its $9-billion budget, with deep cuts to every program except those in President Trump’s priority areas, which include artificial intelligence and quantum information science. NASA, which funds space-based observatories, faces a 25 percent reduction, dropping the agency’s $24.9-billion budget to $18.8 billion. The proposal beefs up efforts to send humans to the moon and to Mars, but the agency’s Science Mission Directorate —home to Mars rovers, the Voyager interstellar probes, the James Webb Space Telescope (JWST), the Hubble Space Telescope, and much more —is looking at a nearly 50 percent reduction, with dozens of missions canceled, turned off or operating on a starvation diet.“It’s an end-game scenario for science at NASA,” says Joel Parriott, director of external affairs and public policy at the American Astronomical Society. “It’s not just the facilities. You’re punching a generation-size hole, maybe a multigenerational hole, in the scientific and technical workforce. You don’t just Cryovac these people and pull them out when the money comes back. People are going to move on.”Adding to the chaos, on Saturday President Trump announced that billionaire entrepreneur and private astronaut Jared Isaacman was no longer his pick for NASA administrator—just days before the Senate was set to confirm Isaacman’s nomination. Initial reports—which have now been disputed—explained the president’s decision as stemming from his discovery that Isaacman recently donated money to Democratic candidates. Regardless of the true reason, the decision leaves both NASA and the NSF, whose director abruptly resigned in April, with respective placeholder “acting” leaders at the top. That leadership vacuum significantly weakens the agencies’ ability to fight the proposed budget cuts and advocate for themselves. “What’s more inefficient than a rudderless agency without an empowered leadership?” Dreier asks.Actions versus WordsDuring his second administration, President Trump has repeatedly celebrated U.S. leadership in space. When he nominated Isaacman last December, Trump noted “NASA’s mission of discovery and inspiration” and looked to a future of “groundbreaking achievements in space science, technology and exploration.” More recently, while celebrating Hubble’s 35th anniversary in April, Trump called the telescope “a symbol of America’s unmatched exploratory might” and declared that NASA would “continue to lead the way in fueling the pursuit of space discovery and exploration.” The administration’s budgetary actions speak louder than Trump’s words, however. Instead of ushering in a new golden age of space exploration—or even setting up the U.S. to stay atop the podium—the president’s budget “narrows down what the cosmos is to moon and Mars and pretty much nothing else,” Dreier says. “And the cosmos is a lot bigger, and there’s a lot more to learn out there.”Dreier notes that when corrected for inflation, the overall NASA budget would be the lowest it’s been since 1961. But in April of that year, the Soviet Union launched the first human into orbit, igniting a space race that swelled NASA’s budget and led to the Apollo program putting American astronauts on the moon. Today China’s rapidprogress and enormous ambitions in space would make the moment ripe for a 21st-century version of this competition, with the U.S. generously funding its own efforts to maintain pole position. Instead the White House’s budget would do the exact opposite.“The seesaw is sort of unbalanced,” says Tony Beasley, director of the NSF-funded National Radio Astronomy Observatory (NRAO). “On the one side, we’re saying, ‘Well, China’s kicking our ass, and we need to do something about that.’ But then we’re not going to give any money to anything that might actually do that.”How NASA will achieve a crewed return to the moon and send astronauts to Mars—goals that the agency now considers part of “winning the second space race”—while also maintaining its leadership in science is unclear.“This is Russ Vought’s budget,” Dreier says, referring to the director of the White House’s Office of Management and Budget (OMB), an unelected bureaucrat who has been notorious for his efforts to reshape the U.S. government by weaponizing federal funding. “This isn’t even Trump’s budget. Trump’s budget would be good for space. This one undermines the president’s own claims and ambitions when it comes to space.”“Low Expectations” at the High FrontierRumors began swirling about the demise of NASA science in April, when a leaked OMB document described some of the proposed cuts and cancellations. Those included both the beleaguered, bloated Mars Sample Return (MSR) program and the on-time, on-budget Nancy Grace Roman Space Telescope, the next astrophysics flagship mission.The top-line numbers in the more fleshed-out proposal are consistent with that document, and MSR would still be canceled. But Roman would be granted a stay of execution: rather than being zeroed out, it would be put on life support.“It’s a reprieve from outright termination, but it’s still a cut for functionally no reason,” Dreier says. “In some ways, [the budget] is slightly better than I was expecting. But I had very low expectations.”In the proposal, many of the deepest cuts would be made to NASA science, which would sink from $7.3 billion to $3.9 billion. Earth science missions focused on carbon monitoring and climate change, as well as programs aimed at education and workforce diversity, would be effectively erased by the cuts. But a slew of high-profile planetary science projects would suffer, too, with cancellations proposed for two future Venus missions, the Juno mission that is currently surveilling Jupiter, the New Horizons mission that flew by Pluto and two Mars orbiters. (The Dragonfly mission to Saturn’s moon Titan would survive, as would the flagship Europa Clipper spacecraft, which launched last October.) NASA’s international partnerships in planetary science fare poorly, too, as the budget rescinds the agency’s involvement with multiple European-led projects, including a Venus mission and Mars rover.The proposal is even worse for NASA astrophysics—the study of our cosmic home—which “really takes it to the chin,” Dreier says, with a roughly $1-billion drop to just $523 million. In the president’s proposal, only three big astrophysics missions would survive: the soon-to-launch Roman and the already-operational Hubble and JWST. The rest of NASA’s active astrophysics missions, which include the Chandra X-ray Observatory, the Fermi Gamma-Ray Space Telescope and the Transiting Exoplanet Survey Satellite (TESS), would be severely pared back or zeroed out. Additionally, the budget would nix NASA’s contributions to large European missions, such as a future space-based gravitational-wave observatory.“This is the most powerful fleet of missions in the history of the study of astrophysics from space,” says John O’Meara, chief scientist at the W. M. Keck Observatory in Hawaii and co-chair of a recent senior review panel that evaluated NASA’s astrophysics missions. The report found that each reviewed mission “continues to be capable of producing important, impactful science.” This fleet, O’Meara adds, is more than the sum of its parts, with much of its power emerging from synergies among multiple telescopes that study the cosmos in many different types, or wavelengths, of light.By hollowing out NASA’s science to ruthlessly focus on crewed missions, the White House budget might be charitably viewed as seeking to rekindle a heroic age of spaceflight—with China’s burgeoning space program as the new archrival. But even for these supposedly high-priority initiatives, the proposed funding levels appear too anemic and meager to give the U.S. any competitive edge. For example, the budget directs about $1 billion to new technology investments to support crewed Mars missions while conservative estimates have projected that such voyages would cost hundreds of billions of dollars more.“It cedes U.S. leadership in space science at a time when other nations, particularly China, are increasing their ambitions,” Dreier says. “It completely flies in the face of the president’s own stated goals for American leadership in space.”Undermining the FoundationThe NSF’s situation, which one senior space scientist predicted would be “diabolical” when the NASA numbers leaked back in April, is also unsurprisingly dire. Unlike NASA, which is focused on space science and exploration, the NSF’s programs span the sweep of scientific disciplines, meaning that even small, isolated cuts—let alone the enormous ones that the budget has proposed—can have shockingly large effects on certain research domains.“Across the different parts of the NSF, the programs that are upvoted are the president’s strategic initiatives, but then everything else gets hit,” Beasley says.Several large-scale NSF-funded projects would escape more or less intact. Among these are the panoramic Vera C. Rubin Observatory, scheduled to unveil its first science images later this month, and the Atacama Large Millimeter/submillimeter Array (ALMA) radio telescope. The budget also moves the Giant Magellan Telescope, which would boast starlight-gathering mirrors totaling more than 25 meters across, into a final design phase. All three of those facilities take advantage of Chile’s pristine dark skies. Other large NSF-funded projects that would survive include the proposed Next Generation Very Large Array of radio telescopes in New Mexico and several facilities at the South Pole, such as the IceCube Neutrino Observatory.If this budget is enacted, however, NSF officials anticipate only funding a measly 7 percent of research proposals overall rather than 25 percent; the number of graduate research fellowships awarded would be cleaved in half, and postdoctoral fellowships in the physical sciences would drop to zero. NRAO’s Green Bank Observatory — home to the largest steerable single-dish radio telescope on the planet — would likely shut down. So would other, smaller observatories in Arizona and Chile. The Thirty Meter Telescope, a humongous, perennially embattled project with no clear site selection, would be canceled. And the budget proposes closing one of the two gravitational-wave detectors used by the LIGO collaboration—whose observations of colliding black holes earned the 2017 Nobel Prize in Physics—even though both detectors need to be online for LIGO’s experiment to work. Even factoring in other operational detectors, such as Virgo in Europe and the Kamioka Gravitational Wave Detector (KAGRA) in Japan, shutting down half of LIGO would leave a gaping blind spot in humanity’s gravitational-wave view of the heavens.“The consequences of this budget are that key scientific priorities, on the ground and in space, will take at least a decade longer—or not be realized at all,” O’Meara says. “The universe is telling its story at all wavelengths. It doesn’t care what you build, but if you want to hear that story, you must build many things.”Dreier, Parriott and others are anticipating fierce battles on Capitol Hill. And already both Democratic and Republican legislators have issued statement signaling that they won’t support the budget request as is. “This sick joke of a budget is a nonstarter,” said Representative Zoe Lofgren of California, ranking member of the House Committee on Science, Space, and Technology, in a recent statement. And in an earlier statement, Senator Susan Collins of Maine, chair of the powerful Senate Committee on Appropriations, cautioned that “the President’s Budget Request is simply one step in the annual budget process.”The Trump administration has “thrown a huge punch here, and there will be a certain back-reaction, and we’ll end up in the middle somewhere,” Beasley says. “The mistake you can make right now is to assume that this represents finalized decisions and the future—because it doesn’t.”

Tools have long been evaluated solely on their utilitarian merits, how efficiently they perform their designated task, how long they last, and how comfortable they feel during extended use. This sterile approach has dominated tool design for generations, yet for those who appreciate the subtle harmony between form and function, it’s simply not enough. There exists a profound hollowness when using tools that assist the hand but leave the mind unmoved, tools that accomplish their purpose without elevating the experience of craftsmanship itself. The Fusion Driver transcends these limitations, transforming the humble screwdriver from a mere implement into an object of desire that commands attention and inspires creativity. Conceived at the intersection of industrial design and functional art, this exquisite tool redefines what a screwdriver can be: a sensory experience that, quite literally, turns the mundane act of driving a screw, into a moment of genuine pleasure. A brief communion between craftsman, tool, and material that elevates the entire process of creation. Designer: Daniel Fajkis Click Here to Buy Now: $89. Hurry, only few left! The tactile experience begins the moment your fingers wrap around the Fusion Driver’s meticulously crafted body. Unlike conventional screwdrivers that prioritize principle at the expense of sensory pleasure, this masterpiece features surfaces that engage your fingertips with subtle variations in texture and haptics. Necessary are such attributes when evolving the average into an obsession, leaving addicted hands defiant to return the tool once a task has been completed. In terms of feel, it’s simply divine. Precision engineering reveals itself in the Fusion Driver’s perfectly tapered weight distribution. The shaft’s mass and center of gravity have been perfectly refined, creating an implement that translates even the slightest wrist movement into efficient rotational force. This linear power transfer lets you feel every subtle variation in resistance as threads engage, providing lossless feedback when precise torque is most important. An experience akin to driving a finely tuned sports car, where every nuance of the road transmits directly to your fingertips. Material selection for the Fusion Driver reflects uncompromising commitment to excellence that would satisfy even the most discerning of collectors. The stainless steel shaft connects to the brass crown via a full-ceramic alumina bearing, each component perfectly balanced and leveled by hand, creating a rotation so smooth it’s bound to be addictive. Interchangeable bits are secured by an N-52 neodymium magnet, providing a connection that feels both secure and trustworthy. This harmonious marriage of premium materials ensures not just durability, but a sensory richness that mass-produced tools simply cannot achieve. The accompanying base transforms storage into a display worthy of the tool itself. Machined from a solid block of stainless steel with twelve powerful N52 neodymium magnets concealed within, it firmly holds up to six 1/4″ hex bits while presenting the Fusion Driver as the objet d’art it truly is. The velvet felt underside protects delicate surfaces, and adds another layer of tactile luxury to the experience. This equally luxurious accessory transforms the Fusion Driver from merely a tool into a statement piece that deserves prominent placement in your workspace. The true value of the Fusion Driver lies not in its ability to perform a task that countless other tools can accomplish, but in how it transforms the task into an experience. Like a fine writing instrument that makes you eager to put pen to paper, the Fusion Driver creates anticipation for work that would otherwise feel mundane. It represents a philosophy that the objects we interact with daily deserve thoughtful design and exquisite execution, and that even the most utilitarian items in our lives can and should bring moments of genuine pleasure and satisfaction. For the discerning collector who understands that true luxury lies in the details, the Fusion Driver represents the pinnacle of design, acknowledging that we are sensory beings who deserve more from our tools than mere functionality. It creates a connection between craftsman and craft that transforms work into pleasure and utility into art, making it an essential addition to any collection of exceptional objects that enhance our experience of the world. Click Here to Buy Now: $89. Hurry, only few left!The post Fusion Driver: The Luxurious Screwdriver Blending Craftsmanship, Beauty and Utility first appeared on Yanko Design.

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Alibaba Group has introduced QwenLong-L1, a new framework that enables large language models (LLMs) to reason over extremely long inputs. This development could unlock a new wave of enterprise applications that require models to understand and draw insights from extensive documents such as detailed corporate filings, lengthy financial statements, or complex legal contracts. The challenge of long-form reasoning for AI Recent advances in large reasoning models (LRMs), particularly through reinforcement learning (RL), have significantly improved their problem-solving capabilities. Research shows that when trained with RL fine-tuning, LRMs acquire skills similar to human “slow thinking,” where they develop sophisticated strategies to tackle complex tasks. However, these improvements are primarily seen when models work with relatively short pieces of text, typically around 4,000 tokens. The ability of these models to scale their reasoning to much longer contexts (e.g., 120,000 tokens) remains a major challenge. Such long-form reasoning requires a robust understanding of the entire context and the ability to perform multi-step analysis. “This limitation poses a significant barrier to practical applications requiring interaction with external knowledge, such as deep research, where LRMs must collect and process information from knowledge-intensive environments,” the developers of QwenLong-L1 write in their paper. The researchers formalize these challenges into the concept of “long-context reasoning RL.” Unlike short-context reasoning, which often relies on knowledge already stored within the model, long-context reasoning RL requires models to retrieve and ground relevant information from lengthy inputs accurately. Only then can they generate chains of reasoning based on this incorporated information.  Training models for this through RL is tricky and often results in inefficient learning and unstable optimization processes. Models struggle to converge on good solutions or lose their ability to explore diverse reasoning paths. QwenLong-L1: A multi-stage approach QwenLong-L1 is a reinforcement learning framework designed to help LRMs transition from proficiency with short texts to robust generalization across long contexts. The framework enhances existing short-context LRMs through a carefully structured, multi-stage process: Warm-up Supervised Fine-Tuning (SFT): The model first undergoes an SFT phase, where it is trained on examples of long-context reasoning. This stage establishes a solid foundation, enabling the model to ground information accurately from long inputs. It helps develop fundamental capabilities in understanding context, generating logical reasoning chains, and extracting answers. Curriculum-Guided Phased RL: At this stage, the model is trained through multiple phases, with the target length of the input documents gradually increasing. This systematic, step-by-step approach helps the model stably adapt its reasoning strategies from shorter to progressively longer contexts. It avoids the instability often seen when models are abruptly trained on very long texts. Difficulty-Aware Retrospective Sampling: The final training stage incorporates challenging examples from the preceding training phases, ensuring the model continues to learn from the hardest problems. This prioritizes difficult instances and encourages the model to explore more diverse and complex reasoning paths. QwenLong-L1 process Source: arXiv Beyond this structured training, QwenLong-L1 also uses a distinct reward system. While training for short-context reasoning tasks often relies on strict rule-based rewards (e.g., a correct answer in a math problem), QwenLong-L1 employs a hybrid reward mechanism. This combines rule-based verification, which ensures precision by checking for strict adherence to correctness criteria, with an “LLM-as-a-judge.” This judge model compares the semanticity of the generated answer with the ground truth, allowing for more flexibility and better handling of the diverse ways correct answers can be expressed when dealing with long, nuanced documents. Putting QwenLong-L1 to the test The Alibaba team evaluated QwenLong-L1 using document question-answering (DocQA) as the primary task. This scenario is highly relevant to enterprise needs, where AI must understand dense documents to answer complex questions.  Experimental results across seven long-context DocQA benchmarks showed QwenLong-L1’s capabilities. Notably, the QWENLONG-L1-32B model (based on DeepSeek-R1-Distill-Qwen-32B) achieved performance comparable to Anthropic’s Claude-3.7 Sonnet Thinking, and outperformed models like OpenAI’s o3-mini and Qwen3-235B-A22B. The smaller QWENLONG-L1-14B model also outperformed Google’s Gemini 2.0 Flash Thinking and Qwen3-32B.  Source: arXiv An important finding relevant to real-world applications is how RL training results in the model developing specialized long-context reasoning behaviors. The paper notes that models trained with QwenLong-L1 become better at “grounding” (linking answers to specific parts of a document), “subgoal setting” (breaking down complex questions), “backtracking” (recognizing and correcting their own mistakes mid-reasoning), and “verification” (double-checking their answers). For instance, while a base model might get sidetracked by irrelevant details in a financial document or get stuck in a loop of over-analyzing unrelated information, the QwenLong-L1 trained model demonstrated an ability to engage in effective self-reflection. It could successfully filter out these distractor details, backtrack from incorrect paths, and arrive at the correct answer. Techniques like QwenLong-L1 could significantly expand the utility of AI in the enterprise. Potential applications include legal tech (analyzing thousands of pages of legal documents), finance (deep research on annual reports and financial filings for risk assessment or investment opportunities) and customer service (analyzing long customer interaction histories to provide more informed support). The researchers have released the code for the QwenLong-L1 recipe and the weights for the trained models. Daily insights on business use cases with VB Daily If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI. Read our Privacy Policy Thanks for subscribing. Check out more VB newsletters here. An error occured.

State-of-the-art models show human-competitive accuracy on AIME, GPQA, MATH-500, and OlympiadBench, solving Olympiad-level problems. Recent multimodal foundation models have advanced benchmarks for disciplinary knowledge and mathematical reasoning. However, these evaluations miss a crucial aspect of machine intelligence: physical reasoning, which requires integrating disciplinary knowledge, symbolic operations, and real-world constraints. Physical problem-solving differs fundamentally from pure mathematical reasoning as it demands models to decode implicit conditions in questions. For example, interpreting “smooth surface” as zero friction coefficient, and maintaining physical consistency across reasoning chains because physical laws remain constant regardless of reasoning trajectories. MLLM shows excellent visual understanding by integrating visual and textual data across various tasks, motivating exploration of its reasoning abilities. However, uncertainty remains regarding whether these models possess genuine advanced reasoning capabilities for visual tasks, particularly in physical domains closer to real-world scenarios. Several LLM benchmarks have emerged to evaluate reasoning abilities, with PHYBench being most relevant for physics reasoning. MLLM scientific benchmarks, such as PhysReason and EMMA, contain multimodal physics problems with figures, however, they include only small physics subsets, which inadequately evaluate MLLMs’ capabilities for reasoning and solving advanced physics problems. Researchers from the University of Hong Kong, the University of Michigan, the University of Toronto, the University of Waterloo, and the Ohio State University have proposed PHYX, a novel benchmark to evaluate the physical reasoning capabilities of foundation models. It comprises 3,000 visually-grounded physics questions, precisely curated across six distinct physics domains: Mechanics, Electromagnetism, Thermodynamics, Wave/Acoustics, Optics, and Modern Physics. It evaluates physics-based reasoning via multimodal problem-solving with three core innovations: (a) 3,000 newly collected questions with realistic physical scenarios requiring integrated visual analysis and causal reasoning, (b) Expert-validated data design covering six fundamental physics domains, and (c) Strict unified three-step evaluation protocols. Researchers designed a four-stage data collection process to ensure high-quality data. The process begins with an in-depth survey of core physics disciplines to determine coverage across diverse domains and subfields, followed by the recruitment of STEM graduate students as expert annotators. They comply with copyright restrictions and avoid data contamination by selecting questions without answers that are immediately available. Moreover, quality control involves a three-stage cleaning process including duplicate detection through lexical overlap analysis with manual review by physics Ph.D. students, followed by filtering the shortest 10% of questions based on textual length, resulting in 3,000 high-quality questions from an initial collection of 3,300. PHYX presents significant challenges for current models, with even the worst-performing human experts achieving 75.6% accuracy, outperforming all evaluated models and showing a gap between human expertise and current model capabilities. The benchmark reveals that multiple-choice formats narrow performance gaps by allowing weaker models to rely on surface-level cues, but open-ended questions demand genuine reasoning and precise answer generation. Comparing GPT-4o’s performance on PHYX to previously reported results on MathVista and MATH-V (both 63.8%), lower accuracy in physical reasoning tasks emphasizes that physical reasoning requires deeper integration of abstract concepts and real-world knowledge, presenting greater challenges than purely mathematical contexts. In conclusion, researchers introduced PHYX, the first large-scale benchmark for evaluating physical reasoning in multimodal, visually grounded scenarios. Rigorous evaluation reveals that state-of-the-art models show limitations in physical reasoning, relying predominantly on memorized knowledge, mathematical formulas, and superficial visual patterns rather than genuine understanding of physical principles. The benchmark focuses exclusively on English-language prompts and annotations, limiting assessment of multilingual reasoning abilities. Also, while images depict physically realistic scenarios, they are often schematic or textbook-style rather than real-world photographs, which may not fully capture the complexity of perception in natural environments. Check out the Paper, Code and Project Page. All credit for this research goes to the researchers of this project. Also, feel free to follow us on Twitter and don’t forget to join our 95k+ ML SubReddit and Subscribe to our Newsletter. Sajjad AnsariSajjad Ansari is a final year undergraduate from IIT Kharagpur. As a Tech enthusiast, he delves into the practical applications of AI with a focus on understanding the impact of AI technologies and their real-world implications. He aims to articulate complex AI concepts in a clear and accessible manner.Sajjad Ansarihttps://www.marktechpost.com/author/sajjadansari/Meta AI Introduces Multi-SpatialMLLM: A Multi-Frame Spatial Understanding with Multi-modal Large Language ModelsSajjad Ansarihttps://www.marktechpost.com/author/sajjadansari/Can LLMs Really Judge with Reasoning? Microsoft and Tsinghua Researchers Introduce Reward Reasoning Models to Dynamically Scale Test-Time Compute for Better AlignmentSajjad Ansarihttps://www.marktechpost.com/author/sajjadansari/NVIDIA AI Introduces AceReason-Nemotron for Advancing Math and Code Reasoning through Reinforcement LearningSajjad Ansarihttps://www.marktechpost.com/author/sajjadansari/Researchers Introduce MMLONGBENCH: A Comprehensive Benchmark for Long-Context Vision-Language Models

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Alibaba Group has introduced QwenLong-L1, a new framework that enables large language models (LLMs) to reason over extremely long inputs. This development could unlock a new wave of enterprise applications that require models to understand and draw insights from extensive documents such as detailed corporate filings, lengthy financial statements, or complex legal contracts. The challenge of long-form reasoning for AI Recent advances in large reasoning models (LRMs), particularly through reinforcement learning (RL), have significantly improved their problem-solving capabilities. Research shows that when trained with RL fine-tuning, LRMs acquire skills similar to human “slow thinking,” where they develop sophisticated strategies to tackle complex tasks. However, these improvements are primarily seen when models work with relatively short pieces of text, typically around 4,000 tokens. The ability of these models to scale their reasoning to much longer contexts (e.g., 120,000 tokens) remains a major challenge. Such long-form reasoning requires a robust understanding of the entire context and the ability to perform multi-step analysis. “This limitation poses a significant barrier to practical applications requiring interaction with external knowledge, such as deep research, where LRMs must collect and process information from knowledge-intensive environments,” the developers of QwenLong-L1 write in their paper. The researchers formalize these challenges into the concept of “long-context reasoning RL.” Unlike short-context reasoning, which often relies on knowledge already stored within the model, long-context reasoning RL requires models to retrieve and ground relevant information from lengthy inputs accurately. Only then can they generate chains of reasoning based on this incorporated information.  Training models for this through RL is tricky and often results in inefficient learning and unstable optimization processes. Models struggle to converge on good solutions or lose their ability to explore diverse reasoning paths. QwenLong-L1: A multi-stage approach QwenLong-L1 is a reinforcement learning framework designed to help LRMs transition from proficiency with short texts to robust generalization across long contexts. The framework enhances existing short-context LRMs through a carefully structured, multi-stage process: Warm-up Supervised Fine-Tuning (SFT): The model first undergoes an SFT phase, where it is trained on examples of long-context reasoning. This stage establishes a solid foundation, enabling the model to ground information accurately from long inputs. It helps develop fundamental capabilities in understanding context, generating logical reasoning chains, and extracting answers. Curriculum-Guided Phased RL: At this stage, the model is trained through multiple phases, with the target length of the input documents gradually increasing. This systematic, step-by-step approach helps the model stably adapt its reasoning strategies from shorter to progressively longer contexts. It avoids the instability often seen when models are abruptly trained on very long texts. Difficulty-Aware Retrospective Sampling: The final training stage incorporates challenging examples from the preceding training phases, ensuring the model continues to learn from the hardest problems. This prioritizes difficult instances and encourages the model to explore more diverse and complex reasoning paths. QwenLong-L1 process Source: arXiv Beyond this structured training, QwenLong-L1 also uses a distinct reward system. While training for short-context reasoning tasks often relies on strict rule-based rewards (e.g., a correct answer in a math problem), QwenLong-L1 employs a hybrid reward mechanism. This combines rule-based verification, which ensures precision by checking for strict adherence to correctness criteria, with an “LLM-as-a-judge.” This judge model compares the semanticity of the generated answer with the ground truth, allowing for more flexibility and better handling of the diverse ways correct answers can be expressed when dealing with long, nuanced documents. Putting QwenLong-L1 to the test The Alibaba team evaluated QwenLong-L1 using document question-answering (DocQA) as the primary task. This scenario is highly relevant to enterprise needs, where AI must understand dense documents to answer complex questions.  Experimental results across seven long-context DocQA benchmarks showed QwenLong-L1’s capabilities. Notably, the QWENLONG-L1-32B model (based on DeepSeek-R1-Distill-Qwen-32B) achieved performance comparable to Anthropic’s Claude-3.7 Sonnet Thinking, and outperformed models like OpenAI’s o3-mini and Qwen3-235B-A22B. The smaller QWENLONG-L1-14B model also outperformed Google’s Gemini 2.0 Flash Thinking and Qwen3-32B.  Source: arXiv An important finding relevant to real-world applications is how RL training results in the model developing specialized long-context reasoning behaviors. The paper notes that models trained with QwenLong-L1 become better at “grounding” (linking answers to specific parts of a document), “subgoal setting” (breaking down complex questions), “backtracking” (recognizing and correcting their own mistakes mid-reasoning), and “verification” (double-checking their answers). For instance, while a base model might get sidetracked by irrelevant details in a financial document or get stuck in a loop of over-analyzing unrelated information, the QwenLong-L1 trained model demonstrated an ability to engage in effective self-reflection. It could successfully filter out these distractor details, backtrack from incorrect paths, and arrive at the correct answer. Techniques like QwenLong-L1 could significantly expand the utility of AI in the enterprise. Potential applications include legal tech (analyzing thousands of pages of legal documents), finance (deep research on annual reports and financial filings for risk assessment or investment opportunities) and customer service (analyzing long customer interaction histories to provide more informed support). The researchers have released the code for the QwenLong-L1 recipe and the weights for the trained models. Daily insights on business use cases with VB Daily If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI. Read our Privacy Policy Thanks for subscribing. Check out more VB newsletters here. An error occured.