Bungie has pulled back the curtain on Marathon with a meaty gameplay deep dive, showing exactly how its cyberpunk-themed PvP extraction shooter will play when it launches on September 23, 2025. From class customization to faction support and the thrill of high-risk, high-reward combat, the game is shaping up to be a uniquely Bungie-flavored evolution of the extraction shooter genre. Players become cybernetic mercenaries known as Runners, humans who’ve ditched their flesh for advanced biosynthetic bodies, each with unique abilities and stats. The core gameplay loop revolves around forming 3-player crews (or going solo), deploying into hostile zones across Tau Ceti IV, scavenging loot, fighting rivals, and extracting before it’s too late. Combat, Classes, and Customization Combat in Marathon is designed to be both tactical and fast-paced, with players customizing loadouts and choosing their Runner class before each deployment. Bungie has already confirmed multiple distinct Runner types: Locus – The frontline “soldier” archetype. Perfect for aggressive players who want to charge into the action and control firefights. Blackbird – A recon-focused Runner. Equipped with tools to detect enemy presence and guide teammates with strategic intel. Glitch – A disruptor class. Geared towards aggressive sabotage and control, Glitch pushes forward to create chaos among enemy lines. Void – The stealth specialist. With active camouflage and smoke deployables, Void thrives in confusion and silence. Each class supports different playstyles and team strategies. And thanks to Bungie’s experience with Destiny, the gunplay looks satisfyingly dynamic and responsive. Factions & Sponsorships Even if you’re wiped out, Marathon ensures players aren’t left behind. Factions in the game can sponsor struggling Runners with fresh gear, encouraging players to jump back in quickly. Building faction reputation may unlock exclusive gear and progression perks over time. Zero to Hero Runs & Loot Economy Much like Escape from Tarkov, if you survive a run, you keep your loot. That loot can then be used to gear up for future runs, creating a strong incentive loop and meaningful progression. If you die, however, everything you brought in can be taken by others. This creates thrilling decision-making moments: Do you push deeper for high-value loot, or extract early with what you have? Endgame and Replayability Seasonal content: The game will launch with Season 1, featuring ongoing story progression and secrets to uncover. Ranked play and community events promise a live-service backbone. Dynamic events, changing weather, and unpredictable enemy players ensure that no two runs ever play out the same. Solo Runners Welcome Unlike many team-focused shooters, Marathon also allows players to opt out of crew fill and run solo. This gives high-skill players the freedom to chase glory alone, with all rewards (and risks) to themselves. You can check out the full gameplay overview below.

Live from TED2025 What’s next for AI? A conversation with OpenAI’s Sam Altman — live at TED2025 Posted by: TED Staff April 12, 2025 at 7:00 am EDT The AI revolution is here to stay, says Sam Altman, the CEO of OpenAI. In a probing, live conversation with head of TED Chris Anderson, Altman discusses the astonishing growth of AI and shows how models like ChatGPT could soon become extensions of ourselves. He also addresses questions of safety, power and moral authority, reflecting on the world he envisions — where AI will almost certainly outpace human intelligence. Watch the full, unedited interview.

Are we unlocking new frontiers in AI reasoning, or simply stretching the limits of token memory without meaningful improvements?Read More

Rob McIlhenney and Charlotte Nicdao in Mythic Quest. Mythic Quest, the Apple TV Plus comedy about a game studio, isn’t getting a fifth season, reports Variety. And, like a developer issuing a farewell patch, Apple is updating the final episode of season four with a new “goodbye” ending next week. Variety published a statement from Mythic Quest’s executive producers: “Endings are hard. But after four incredible seasons, ‘Mythic Quest’ is coming to a close,” said series executive producers Megan Ganz, David Hornsby, and Rob McElhenney. “We’re so proud of the show and the world we got to build—and deeply grateful to every cast and crew member who poured their heart into it. To all our fans, thank you for playing with us. To our partners at Apple, thank you for believing in the vision from the very beginning. Because endings are hard, with Apple’s blessing we made one final update to our last episode—so we could say goodbye, instead of just game over.” The news comes just weeks after a report that Apple has been losing $1 billion a year on TV Plus. The show just wrapped up its fourth season and released a four-episode spinoff called Side Quest on March 26th. One of the early Apple TV Plus shows, Mythic … Read the full story at The Verge.

Multimodal AI enables machines to process and reason across various input formats, such as images, text, videos, and complex documents. This domain has seen increased interest as traditional language models, while powerful, are inadequate when confronted with visual data or when contextual interpretation spans across multiple input types. The real world is inherently multimodal, so systems aiming to assist in real-time tasks, analyzing user interfaces, understanding academic materials, or interpreting complex scenes require intelligence that functions beyond textual reasoning. Newer models are now being developed to simultaneously decode language and vision cues to approach tasks with improved contextual awareness, reasoning depth, and adaptability to different data input forms. A limitation in multimodal systems today lies in their inability to process long contexts efficiently and to generalize across high-resolution or diverse input structures without compromising performance. Many open-source models limit the input to a few thousand tokens or demand excessive computational resources to maintain performance at scale. These constraints result in models that may perform well on standard benchmarks but struggle with real-world applications that involve complex, multi-image inputs, extended dialogues, or academic tasks like OCR-based document analysis and mathematical problem-solving. There’s also a gap in reasoning ability, particularly long-horizon thinking, which prevents current systems from handling tasks that require step-by-step logic or deep contextual alignment between different data modalities. Previous tools have attempted to address these challenges but often fell short in scalability or flexibility. The Qwen2.5-VL series and Gemma-3 models, while notable for their dense architectures, lack built-in support for reasoning through longer chains of thought. Models like DeepSeek-VL2 and Aria adopted mixture-of-experts (MoE) strategies but had fixed vision encoders that restricted their ability to adapt to various resolutions and forms of visual input. Also, these models typically supported only short context windows, 4K tokens in DeepSeek-VL2, and had limited success in complex OCR or multi-image scenarios. As such, most existing systems failed to balance low resource consumption with the ability to tackle tasks involving long context and diverse visual data. Researchers at Moonshot AI introduced Kimi-VL, a novel vision-language model utilizing an MoE architecture. This system activates only 2.8 billion parameters in its decoder, significantly lighter than many competitors while maintaining powerful multimodal capabilities. The two released models based on this architecture on Hugging Face are Kimi-VL-A3B-Thinking and Kimi-VL-A3B-Instruct. It incorporates a native-resolution visual encoder named MoonViT and supports context windows of up to 128K tokens. The model has three integrated components: the MoonViT encoder, an MLP projector for transitioning visual features to language embeddings, and the Moonlight MoE decoder. Researchers further developed an advanced version, Kimi-VL-Thinking, designed specifically for long-horizon reasoning tasks through chain-of-thought supervised fine-tuning and reinforcement learning. Together, these models aim to redefine efficiency benchmarks in vision-language reasoning. The architectural innovation in Kimi-VL lies in its adaptability and processing capability. MoonViT processes high-resolution images in their original form, eliminating the need for sub-image fragmentation. To ensure spatial consistency across varied image resolutions, the model uses interpolated absolute positional embeddings combined with two-dimensional rotary positional embeddings across both height and width. These design choices allow MoonViT to preserve fine-grained detail even in large-scale image inputs. Outputs from the vision encoder are passed through a two-layer MLP that uses pixel shuffle operations to downsample spatial dimensions and convert features into LLM-compatible embeddings. On the language side, the 2.8B activated parameter MoE decoder supports 16B total parameters and integrates seamlessly with visual representations, enabling highly efficient training and inference across different input types. The entire training process used an enhanced Muon optimizer with weight decay and ZeRO-1-based memory optimization for handling the large parameter count. The training data composition reflects a focus on diverse multimodal learning. Starting with 2.0T tokens for ViT training using image-caption pairs, the team added another 0.1T to align the encoder with the decoder. Joint pre-training consumed 1.4T tokens, followed by 0.6T in cooldown and 0.3T in long-context activation, totaling 4.4T tokens. These stages included academic visual datasets, OCR samples, long video data, and synthetic mathematical and code-based QA pairs. For long-context learning, the model was progressively trained to handle sequences from 8K up to 128K tokens, using RoPE embeddings extended from a base frequency of 50,000 to 800,000. This allowed the model to maintain a token recall accuracy of 100% up to 64K tokens, with a slight drop to 87.0% at 128K, still outperforming most alternatives. Kimi-VL demonstrated strong results across a range of benchmarks. On the LongVideoBench, it scored 64.5; on MMLongBench-Doc, it achieved 35.1; and on the InfoVQA benchmark, it led with 83.2. On ScreenSpot-Pro, which tests understanding of UI screens, it scored 34.5. The Kimi-VL-Thinking variant excelled in reasoning-intensive benchmarks like MMMU (61.7), MathVision (36.8), and MathVista (71.3). For agent tasks such as OSWorld, the model matched or exceeded performance from larger models like GPT-4o while activating significantly fewer parameters. Its compact design and strong reasoning capabilities make it a leading candidate among open-source multimodal solutions. Some Key Takeaways from the Research on Kimi-VL: Kimi-VL activates only 2.8B parameters during inference, ensuring efficiency without sacrificing capability.  MoonViT, its vision encoder, natively processes high-resolution images, improving clarity in tasks like OCR and UI interpretation.   The model supports up to 128K context tokens, achieving 100% recall up to 64K and 87.0% accuracy at 128K on text/video tasks.   Kimi-VL-Thinking scores 61.7 on MMMU, 36.8 on MathVision, and 71.3 on MathVista, outperforming many larger VLMs.   It scored 83.2 on InfoVQA and 34.5 on visual tasks on ScreenSpot-Pro, showcasing its precision in perception-based evaluations.   Total pre-training involved 4.4T tokens across text, video, document, and synthetic multimodal data.  Optimization was done using a customized Muon optimizer with memory-efficient strategies like ZeRO-1.  Joint training ensured seamless visual and language feature integration while preserving core language capabilities. Check out Twitter and don’t forget to join our 85k+ ML SubReddit. Sana HassanSana Hassan, a consulting intern at Marktechpost and dual-degree student at IIT Madras, is passionate about applying technology and AI to address real-world challenges. With a keen interest in solving practical problems, he brings a fresh perspective to the intersection of AI and real-life solutions.Sana Hassanhttps://www.marktechpost.com/author/sana-hassan/Google AI Introduce the Articulate Medical Intelligence Explorer (AMIE): A Large Language Model Optimized for Diagnostic Reasoning, and Evaluate its Ability to Generate a Differential DiagnosisSana Hassanhttps://www.marktechpost.com/author/sana-hassan/Balancing Accuracy and Efficiency in Language Models: A Two-Phase RL Post-Training Approach for Concise ReasoningSana Hassanhttps://www.marktechpost.com/author/sana-hassan/RoR-Bench: Revealing Recitation Over Reasoning in Large Language Models Through Subtle Context ShiftsSana Hassanhttps://www.marktechpost.com/author/sana-hassan/T* and LV-Haystack: A Spatially-Guided Temporal Search Framework for Efficient Long-Form Video Understanding

If Escape From Tarkov is the PUBG of the extraction shooter, boldly striking out into a new genre despite humble origins, then Bungie’s upcoming shooter, Marathon, is the big-budget refinement that could become the emerging category’s Fortnite. Coming from a studio with a pedigree for first-person shooters that’s virtually unmatched, and which wields the resources to take a good idea and reimagine it in a disturbing, beautifully detailed sci-fi universe, it’s easy to see how Bungie just might pull it off. But with a concept that’s so inherently geared towards the hardcore PvP crowd and such a clear departure from what they’ve done in the past, it’s certainly not without risks, even if they do manage to capture the usual FPS magic they’re known for. Even so, after spending the better part of two days sliding and gunning my way through Tau Ceti’s relentless ruins, I was hooked, and can’t wait to group up with my most tryhard friends and ruin people’s day.Marathon is an extraction shooter where you and two friends are dropped into an extremely unmerciful wasteland to battle deadly enemies, including rival players, all in the name of loot and glory. If you’ve played Escape From Tarkov, then you’ll be familiar with this PvPvE framing, which forces you to ante up any loot you bring along with you, as it’s permanently lost if you fail to exfiltrate before time runs out or your team is killed. Those distressingly high stakes immediately got my blood pumping and made each match feel like the most important one so far. After fighting tooth and nail against surprisingly challenging AI enemies and ruthless fellow Runners to get a high-level rifle that could down most enemies in two hits, I wasn’t about to let my guard down for even a moment and risk losing it all to some unworthy foe.Marathon - Gameplay ScreenshotsIf, like me, you bemoaned the quiet, undignified death of PvP’s relevance in Destiny 2, then you might share my eagerness to see Bungie focused squarely on a competitive experience here, and everything I played so far indicates they’ve still more than got what it takes to pull it off. Weapons and movement feels crisp, fluid, and incredibly satisfying in that rarified way that few studios can achieve, and the added element of hero shooter-like abilities, like one that lets you blast enemies with kinetic energy, sending them flying away, gives that extra kick you need to feel like a supernaturally powered badass. Marathon almost immediately feels fantastic to play, and as soon as I took out a robot who took a potshot at me from the window of a nearby building and tried out his weapon for myself on a nearby enemy, it was love at first snipe.This gets even better once you start to familiarize yourself with the class-based Runners available to you (four of which I got to play extensively, out of the six that will be available at launch). Glitch is all about mobility and controlling positioning, and has abilities that help her move fast, jump up to areas others can’t reach, and push enemies away with a blast of kinetic energy. Locus is a tanky soldier who can pull up a shield to block incoming fire and launch a flurry of homing missiles at enemies, while Blackbird is all about intel and using her gadgets to ping any enemies within an area for the whole team, and sending out little robotic drones to seek and destroy nearby enemies. My personal favorite though, was Void: a deft ninja who can turn invisible for a time and deploy massive clouds of smoke to lose and confuse the enemy within. Each of these characters offers a completely different way to play Marathon, and by coordinating with my teammates, we were able to put together a balanced team of Runners who complemented one another’s strengths and helped make up for any weaknesses.PlayThat came in handy as we explored dark and forgotten places throughout Marathon’s maps, which were filled with deadly creatures, like a species of giant ticks who ran at me frantically, and surprisingly accurate robots who were much, much better at surrounding and overwhelming players than the vast majority of AI enemies found in other games. Fighting these foes was no joke, but the loot was well worth it, whether it was finding a backpack that allowed for more storage space, a weapon that gave me better odds against a rival crew, or just some consumable healing items I could rely upon in a pinch. Some of the extremely rare and powerful items, of which my team only found a handful during our matches, changed things up quite significantly, like a backpack that turned the user invisible whenever they interacted with a container while wearing it – extremely useful in a game where you’re being hunted by everyone else on the map.Marathon provided me with tons of memorable moments of elation after a narrow victory and the occasionally sour taste of defeat.“Marathon’s tensest moments, though, happen when you encounter enemy crews and inevitably clash in a bid to claim one another’s loot. Maybe we’d spot a crew in the distance and stealthily follow them – waiting for the right moment to stage an ambush – or sometimes we’d hear other players in the distance fighting off security forces and decide to go catch them when their guard was down, scooping up whatever goodies they were about to claim. And, of course, sometimes we found ourselves on the receiving end of that ruthless equation and had to scramble as one of us was downed by sudden gunfire from a rival team. These moments are when the PvP magic really shines and leaves you with stories to tell your friends, like the time my group got entangled in a nine-player free-for-all and came out unscathed thanks to some well-timed smoke grenades, or another time where both of my teammates were killed, but we came back to win it thanks to a clutch revive the other team didn’t see coming. Just like the most epic, heart-pounding moments of a battle royale match, Marathon provided me with tons of memorable moments of elation after a narrow victory and the occasionally sour taste of defeat.PlayOnce we’d gotten our spoils or completed an objective or two, it was time to haul ass to the exfiltration beacon and stand near it long enough to be pulled out of the firefight and returned safely back to our base in outer space with our bloody bounty. But lighting that beacon also lets everyone else on the map know you’re trying to make off with your goods, via a giant blue light that shoots up into the sky for a prolonged period of time. This makes your last moments in the area pretty intense, or, if you’re like me, offers some exciting opportunities to kill a group right as they’re preparing to leave and benefit from all of their hard work.The downside of Marathon’s “to the victors go the spoils” reward system is that in a pretty short amount of time, those with skill will enter fresh matches with gear that easily outmatches less skilled players, making them even less likely to be able to pull off a victory. Bungie’s gone out of its way to try and make it so a poorly geared squad still stands a chance against those with better loot if they play well, and you can always simply employ the strategy of avoiding other players until you’ve got a few extractions and some at least decent loot under your belt, but even so, it seems like it could get pretty demoralizing in short order if you run into a bout of bad luck and start to feel like you’ll never catch a break.PlayI was lucky enough to win the vast majority of my matches, and after the initial hump of stockpiling some basic equipment in my vault and unlocking a few permanent upgrades to make the going easier, I felt like even if I fell in battle and lost all my best stuff, at least I’d have some adequate backup gear to recover with. But that won’t be the case for everyone, and I wonder how big the chunk of players will be who simply run into a brick wall in their first few hours, log off, and never return. At least in Escape from Tarkov you’ve got your Kappa case to hold your most valuable items and protect them from being lost on the battlefield – Marathon has none of those safety measures, instead opting for a system where you’re given a sort of pity loadout, called a sponsored package, which sets you up with some basic gear after you’ve been quite literally bled dry. This definitely helps ice the wound of a string of bad luck, but might not exactly be the most comforting consolation prize.I wonder how big the chunk of players will be who simply run into a brick wall in their first few hours, log off, and never return.“To be clear, it very well might be a good thing that Marathon seemingly refuses to compromise on its dedication to ensuring each match has real, keenly felt stakes, even if it turns some folks off from it – I certainly count myself among those who enjoy a bit of sweatiness and quite liked how invested I was forced to be every single time I deployed onto a new map. But they definitely risk a situation where it feels like the rich are getting richer and the poor are getting poorer, to the detriment of those who are on the unfortunate side of that brutal equation. Time will tell before we know how that will shake out, but even in my short time with it, I already felt a bit of friction emanating from those on the server who kept dying repeatedly and felt like they were getting outclassed by those with quality loadouts.Marathon PlayStation Showcase 2023 TrailerOne thing that could be the key to preventing people from falling off is Marathon’s ongoing meta progression system, which I found surprisingly engaging in my time with it. Even though your loot is probably destined to be stripped off your still-warm corpse and stuffed into some rude opponent’s backpack, you can at least comfort yourself in the various perk trees and ongoing quests you’ll find in the menus between rounds of play. You might not have made it back alive from your most recent escapade, but maybe you were able to loot some chests from a specific part of the map or complete some other objective in your quest log before you went cold. In that case, you can at least still make your way through the faction questlines and upgrade systems, which unlock new bits of story and improve your reputation with each of the three groups who have taken a marked interest in your progress. Doing so grants you perks that make the going a bit easier, like those that let you buy slightly better gear from the vendor, in case you find all your existing loot confiscated by the enemy.Elsewhere, despite some clear indications that some kind of plot will be included in Marathon even though multiplayer mayhem is plainly the focus, it’s still unclear to me how much that story will matter or deliver something that makes me want to progress, beyond just finding better weapons to use. As a mercenary called a Runner, your consciousness is repeatedly uploaded to a synthetic body and sent into dangerous territory in search of riches. Beyond this basic premise and some really cool details, like how your artificial bodies are sewn together out of synth silk by robotic caterpillars, there wasn’t a whole lot to go off of, and I could easily see this not being an area of particular focus for Marathon. That said, the world Bungie has built is every bit as beautiful as it is creepy and dystopian, and there are moments where they satisfyingly hint at the events of the Marathon trilogy from the ‘90s. But Bungie also told me that they’re not entirely following the events of the original games, and what little I saw did only slightly more than pay lip service to the world.There’s the question of whether Bungie will add enough content at a fast enough pace to appease ever-ravenous players.“Another question is just how much support and longevity we can expect from Marathon. Bungie has earned its reputation for mastering the live-service model over the past decade of Destiny, and shows no signs of letting players down here, as they told me they already have plans to support Marathon with new maps, weapons, characters, and more as they develop it, but as always, there’s the question of whether they’ll do so at a fast enough pace to appease ever-ravenous players. And importantly, Marathon also plans to use the seasonal reset model found in the likes of Diablo 4, where you’ll be stripped of your loot and progress at the end of each season to do so again with new content and a new meta to mix things up. While I could definitely see this working for Marathon, there’s always the question of how they’ll manage to incentivize making that arduous climb from scratch each season.After playing enough Marathon to get a pretty good feel for it in its alpha state, I’m already fairly confident it’ll scratch the Bungie PvP itch that’s been left unscratched for a number of years now. Here’s hoping they’ll knock it out of the park when it comes to PC and consoles in September.

Welcome to Indie App Spotlight. This is a weekly 9to5Mac series where we showcase the latest apps in the indie app world. If you’re a developer and would like your app featured, get in contact. VoicePen is a native app for iPhone, iPad, Mac, and Apple Vision Pro that transcribes your lectures, meetings, and any sort of speech into clear notes. All of the transcriptions run through OpenAI’s Whisper model, making it accurate and fast. It also offers more than just meeting transcriptions, and can be used for YouTube videos and other online content. Top features VoicePen supports recordings up to 2 hours, allowing you to transcribe pretty long meetings and classes. It even supports background recording, so you don’t need to have the app open the whole time. Afterwards, recordings are transcribed through OpenAI’s Whisper model, allowing for precise transcriptions. There’s also options for rewriting the transcription, allowing you to adjust its style for your tastes. There’s also folders for easy organization, and the app supports iCloud Sync, allowing for your same notes to be available across all of your Apple devices. VoicePen also has home screen widgets, live activities, and app intents with Siri. And, as mentioned earlier – it also supports uploads of online content. You can transcribe podcasts, YouTube videos, zoom recordings, and whatever else floats your boat. VoicePen is available for free on the App Store for devices running iOS 17, macOS Sonoma 14.0, or visionOS 1 and later. The free app is fairly capable, but for the full experience, including unlimited recordings and summarizations, you’ll need the premium subscription, which is available for $9.99/month or $44.99/year. Both plans offer free trials. My favorite Apple accessories on Amazon: Follow Michael: X/Twitter, Bluesky, Instagram Add 9to5Mac to your Google News feed.  FTC: We use income earning auto affiliate links. More.You’re reading 9to5Mac — experts who break news about Apple and its surrounding ecosystem, day after day. Be sure to check out our homepage for all the latest news, and follow 9to5Mac on Twitter, Facebook, and LinkedIn to stay in the loop. Don’t know where to start? Check out our exclusive stories, reviews, how-tos, and subscribe to our YouTube channel

Sam Altman's grand vision of democratizing artificial intelligence seems, per a new study, to include an ugly asterisk: it's monetized, at least in part, by spambots filling the web with AI-generated garbage.According to the cybersecurity firm SentinelOne, search engine optimization (SEO) scammers operated undetected for months using what company's experts are calling "AkiraBot," a "modular and sophisticated" tool that was able to bypass CAPTCHAs and other spam detection filters with ease. As SentinelOne explained, the bot was named not for any affiliation with the ransomware group Akira, but because that name, which means "bright" in Japanese, is used in a lot of the creators' affiliated domains.AkiraBot's bottom line seems to be directing traffic to its dubious SEO scheme — and with GPT-4o-mini, that process seems to have been automated at scale. Having attempted to spam roughly 420,000 sites and successfully getting its trash through to some 80,000, the humans behind AkiraBot were almost certainly paying for access to OpenAI's API — and we've reached out to the Altman-run company to confirm.The bot's chief targets, per SentinelOne's investigation, were small and medium-sized businesses — and specifically, the contact forms and chat widgets on those companies' websites. Using GPT-4o-mini to craft templates based on whichever type of contact module was at play, the spammers customized unique messages for each website in ways that got around spam filters at least part of the time.Starting with instructions that tell OpenAI's most cost-efficient advanced model to act like a "helpful assistant that generates marketing messages," the bots' creators operated for months before SentinelOne got wise to its spam scam.In one example of an AkiraBot message from a targeted candle company's comments section, a phony customer service rep named "Megan" shilled SEO services that were, per angry reviews left on the "Akira" trustpilot page, nonexistent."My name is Megan, from The Akira Team — I just noticed your website through your Entireweb Website Listing, and wanted to get in touch with you right away," the spam message reads. "We have a special offer for your website today, and that is 1st Page Rankings in all major search engines (That’s Google, Yahoo and Bing) + social media and video commercial advertising starting at just $29.99 which I am ABSOLUTELY certain will benefit your website and business, by bringing you LOTS of new customers, very very quickly."Upon discovering and analyzing these bots and their output, SentinelOne's researchers alerted OpenAI — and to the company's credit, it immediately investigated and ultimately disabled the creators' account.Still, it managed to run in earnest between September 2024 and February 2025, when AkiraBot got caught — and there's no way to know how long its creators, who weren't named, paid OpenAI for access to its API.Share This Article

Hints and answers for the NYT Connections: Sports Edition puzzle, No. 202, for April 13.

In Brief Posted: 1:15 PM PDT · April 12, 2025 Image Credits:Apple TV+ Apple’s ‘Mythic Quest’ is ending with an updated Season 4 finale “Mythic Quest,” the Apple TV+ workplace comedy about the making of a popular online roleplaying game, is ending after four seasons. A new version of the show’s fourth season finale — first aired on March 26, and now the de facto series finale — will premiere on Apple TV+ next week, with a new ending. Longtime TV fans are used to the disappointment of a prematurely canceled show, and it can feel like a relief when the creators get to wrap up their story with an abbreviated final season, a TV movie, or even extra scenes added to an already-filmed episode. Even more unusual is what Apple is doing here, with a new ending getting edited into an episode that already aired. In a statement provided to Variety, executive producers Megan Ganz, David Hornsby, and Rob McElhenney said that “with Apple’s blessing we made one final update to our last episode — so we could say goodbye, instead of just game over.” Topics