--- title: The Robotics Breakthrough Everyone Has Been Waiting For description: Robotics startup Generalist AI released Gen 1.5, a robot foundation model demonstrating in-context learning: shown a few seconds of a demonstration (video,... canonical: https://daily.dev/posts/the-robotics-breakthrough-everyone-has-been-waiting-for-chxhb0dbw twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: The Robotics Breakthrough Everyone Has Been Waiting For | daily.dev og:description: Robotics startup Generalist AI released Gen 1.5, a robot foundation model demonstrating in-context learning: shown a few seconds of a demonstration (video,... og:url: https://daily.dev/posts/the-robotics-breakthrough-everyone-has-been-waiting-for-chxhb0dbw og:image: https://api.daily.dev/og/posts/CHxhB0dbW.png og:image:alt: The Robotics Breakthrough Everyone Has Been Waiting For og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # The Robotics Breakthrough Everyone Has Been Waiting For **[bycloud](https://daily.dev/sources/bycloud)** · 15 min read · 3 upvotes · 0 comments ## Summary Robotics startup Generalist AI released Gen 1.5, a robot foundation model demonstrating in-context learning: shown a few seconds of a demonstration (video, simulation, or even bare human hands), the robot can immediately attempt a new manipulation task without any weight updates. Across 10 held-out manipulation tasks never seen in training, zero-shot success averages 59%, jumping to 83% after just 10 gradient steps on ~5 minutes of demonstration data (weights change by less than 0.15%). The model also shows emergent behaviors like using a banana as a brush, freeing a stuck Lego piece with its other hand, sorting blocks by color unprompted, and recovering from a knocked-over cup mid-task. Pre-training over 8+ months progressively reduced the gradient steps needed for new tasks from hundreds down to zero. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is Gen 1.5 from Generalist AI and how does its in-context learning work for robots? Gen 1.5 is a robot foundation model from Generalist AI, released around August 2026, that can perform new manipulation tasks after being shown just a 3 to 12 second demonstration placed in its context window, without any weight updates. It has roughly 30 seconds of video memory plus proprioceptive and language inputs, processing actions at 100 hertz, and functions like next-action prediction rather than next-token prediction. _Developers tracking foundation model breakthroughs can follow releases like this one on daily.dev._ ### How well does Gen 1.5 perform on new manipulation tasks without fine-tuning versus with minimal fine-tuning? Zero-shot, using only a physical demonstration placed in context with no gradient updates, Gen 1.5 succeeds on average 59% across 10 held-out manipulation tasks. After fine-tuning on just 5 minutes of demonstrations (about 50 examples) for only 10 gradient steps, average success rises to 83%, with sweeping-with-a-brush jumping from 37% to 99% and jar-opening from 60% to 94.5%. _Engineers evaluating model efficiency trade-offs can compare benchmarks like these on daily.dev._ ### What emergent behaviors has Gen 1.5 shown that were not explicitly trained? Gen 1.5 has displayed physical common sense beyond its training data, such as using a banana as an improvised brush, switching strategies entirely to scoop with a dustpan instead of sweeping, sorting multiple blocks by color when only trained to place one block in a bowl, freeing a Lego piece accidentally stuck to its gripper using its other hand, and moving a piece of paper covering a bowl before completing a task. _Anyone watching emergent AI capabilities unfold can keep up with reports like this on daily.dev._ --- Tags: [#ai](https://daily.dev/tags/ai), [#deep-learning](https://daily.dev/tags/deep-learning), [#robotics](https://daily.dev/tags/robotics) [View this post on daily.dev](https://daily.dev/posts/the-robotics-breakthrough-everyone-has-been-waiting-for-chxhb0dbw) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"The Robotics Breakthrough Everyone Has Been Waiting For","url":"https://daily.dev/posts/the-robotics-breakthrough-everyone-has-been-waiting-for-chxhb0dbw","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/the-robotics-breakthrough-everyone-has-been-waiting-for-chxhb0dbw"},"datePublished":"2026-09-02T20:54:24.007Z","dateModified":"2026-09-02T20:54:48.088Z","description":"Robotics startup Generalist AI released Gen 1.5, a robot foundation model demonstrating in-context learning: shown a few seconds of a demonstration (video,...","image":"https://i.ytimg.com/vi/cqwKceUSZ5Q/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/cqwKceUSZ5Q/sddefault.jpg","isAccessibleForFree":true,"articleSection":"bycloud","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"bycloud","logo":"https://media.daily.dev/image/upload/s--UUgAibgb--/f_auto/v1712315122/logos/bycloud","url":"https://daily.dev/sources/bycloud"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/the-robotics-breakthrough-everyone-has-been-waiting-for-chxhb0dbw","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":3},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai,deep-learning,robotics","timeRequired":"PT15M","video":{"@type":"VideoObject","name":"The Robotics Breakthrough Everyone Has Been Waiting For","description":"Robotics startup Generalist AI released Gen 1.5, a robot foundation model demonstrating in-context learning: shown a few seconds of a demonstration (video,...","thumbnailUrl":"https://i.ytimg.com/vi/cqwKceUSZ5Q/sddefault.jpg","uploadDate":"2026-09-02T20:54:24.007Z","duration":"PT15M","url":"https://api.daily.dev/r/CHxhB0dbW","embedUrl":"https://www.youtube.com/embed/cqwKceUSZ5Q"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"bycloud","item":"https://daily.dev/sources/bycloud"},{"@type":"ListItem","position":3,"name":"The Robotics Breakthrough Everyone Has Been Waiting For"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/the-robotics-breakthrough-everyone-has-been-waiting-for-chxhb0dbw#faq","mainEntity":[{"@type":"Question","name":"What is Gen 1.5 from Generalist AI and how does its in-context learning work for robots?","acceptedAnswer":{"@type":"Answer","text":"Gen 1.5 is a robot foundation model from Generalist AI, released around August 2026, that can perform new manipulation tasks after being shown just a 3 to 12 second demonstration placed in its context window, without any weight updates. It has roughly 30 seconds of video memory plus proprioceptive and language inputs, processing actions at 100 hertz, and functions like next-action prediction rather than next-token prediction. Developers tracking foundation model breakthroughs can follow releases like this one on daily.dev."}},{"@type":"Question","name":"How well does Gen 1.5 perform on new manipulation tasks without fine-tuning versus with minimal fine-tuning?","acceptedAnswer":{"@type":"Answer","text":"Zero-shot, using only a physical demonstration placed in context with no gradient updates, Gen 1.5 succeeds on average 59% across 10 held-out manipulation tasks. After fine-tuning on just 5 minutes of demonstrations (about 50 examples) for only 10 gradient steps, average success rises to 83%, with sweeping-with-a-brush jumping from 37% to 99% and jar-opening from 60% to 94.5%. Engineers evaluating model efficiency trade-offs can compare benchmarks like these on daily.dev."}},{"@type":"Question","name":"What emergent behaviors has Gen 1.5 shown that were not explicitly trained?","acceptedAnswer":{"@type":"Answer","text":"Gen 1.5 has displayed physical common sense beyond its training data, such as using a banana as an improvised brush, switching strategies entirely to scoop with a dustpan instead of sweeping, sorting multiple blocks by color when only trained to place one block in a bowl, freeing a Lego piece accidentally stuck to its gripper using its other hand, and moving a piece of paper covering a bowl before completing a task. Anyone watching emergent AI capabilities unfold can keep up with reports like this on daily.dev."}}]} ```