Breakthrough AI to solve the world's biggest problems. › Join us: http://allenai.org/careers › Get our newsletter: https://share.hsforms.com/1uJkWs5aDRHWhiky3aHooIg3ioxmhttps://bsky.app/profile/ai2.bsky.socialhttps://bsky.app/profile/ai2.bsky.social/post/3mviql3olk22nHow should future scientists learn to interrogate AI tools for discovery? Read how UW students put AutoDiscovery to the test as part of an academic challenge earlier this year, then try the tool for yourself—credits are now extended through Dec. 31. 🧵 buff.ly/fXoQxxY14 Sep 2026 18:30 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mviql3olk22nhttps://bsky.app/profile/ai2.bsky.social/post/3mv3wv46gq724Training an LLM by showing it answers people prefer – and ones they don’t – can improve it while quietly worsening behaviors. Goodfire AI used Ai2’s open post-training stack to predict how a full training run would change responses to different prompts. 🧵 buff.ly/RuJxLM109 Sep 2026 16:19 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mv3wv46gq724https://bsky.app/profile/ai2.bsky.social/post/3muztlrwqti2gWe're at #ECCV2026 with papers & talks across the conference. Come say hello and learn about our latest research!08 Sep 2026 20:14 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3muztlrwqti2ghttps://bsky.app/profile/ai2.bsky.social/post/3mupnqr5ljw2qML emulators mimic climate processes faster than traditional models. Next is coupling atmosphere & ocean emulators so their predictions feed into each other as the simulation runs. With the E3SM team, we built a system that does that: SamudrACE-E3SMv3. 🧵 buff.ly/9ShSTCM04 Sep 2026 19:03 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mupnqr5ljw2qhttps://bsky.app/profile/ai2.bsky.social/post/3muifsmwq2n2yDo LLM safety & capability evals measure what they claim to? We built BenchMIRT to audit them + see which model abilities their Qs actually test. On BBQ, a social-bias eval, it found the Qs distinguished models more by reasoning ability than safety. 🧵 buff.ly/bTcvqJf01 Sep 2026 21:52 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3muifsmwq2n2yhttps://bsky.app/profile/ai2.bsky.social/post/3muhnldyvwg2zAt an event on August 27, we brought together AI researchers, scientists, & medical practitioners to explore what AI needs to do better to meaningfully advance science. Five ideas kept coming up. 🧵01 Sep 2026 14:39 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3muhnldyvwg2zhttps://bsky.app/profile/ai2.bsky.social/post/3mu2w2susev2kAI’s biggest role in science may not be answering questions. It may be helping scientists find which questions are worth asking. At Providence Swedish, AutoDiscovery surfaced an unexpected immune signal in a heavily studied cancer dataset—and follow-up research confirmed it. 🧵 buff.ly/LTacUQl27 Aug 2026 13:06 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mu2w2susev2khttps://bsky.app/profile/ai2.bsky.social/post/3mtyxnp2bew2kA Thai research team adapted our Dolma data-curation toolkit to build Mangosteen, a 47B-token corpus for Thai LLMs. They used Dolma to filter widely used web datasets into a smaller corpus that improved Thai LLM performance despite using less data. 🧵 buff.ly/fvLSIru26 Aug 2026 18:29 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mtyxnp2bew2khttps://bsky.app/profile/ai2.bsky.social/post/3mtmjmmad3l2pWhat kinds of training data shape different model capabilities? A Georgia Tech team used our fully open model flow to trace Olmo’s performance on social/general reasoning and social-science/STEM knowledge tests back to the types of text it trained on. 🧵 buff.ly/cQEzbUn21 Aug 2026 19:46 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mtmjmmad3l2phttps://bsky.app/profile/ai2.bsky.social/post/3mtevy2m3dn2eA model can sound like it knows a drug—even when it doesn’t. Researchers at @utaustin.bsky.social, @northeasternu.bsky.social, & @mdanderson.bsky.social found LLMs often lack drug-specific knowledge and instead lean on morphology. They used Olmo to trace why. 🧵 buff.ly/PuyfnCq18 Aug 2026 19:06 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mtevy2m3dn2ehttps://bsky.app/profile/ai2.bsky.social/post/3msixvpixoj27Today we're introducing a preview of TutorMoments, a framework that measures whether AI tutors can make one of the hardest calls in teaching: when to step in and help a student, & when to hold back and let them do the heavy thinking. 🧵07 Aug 2026 16:26 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3msixvpixoj27https://bsky.app/profile/ai2.bsky.social/post/3msg4ymhood2bWe're expanding our partnership with @hf.co to accelerate open science. Our storage on the Hub is roughly tripling to ~2 petabytes, & our downloads now run at high speed—even for our largest datasets & multi-checkpoint models. 🧵06 Aug 2026 13:19 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3msg4ymhood2bhttps://bsky.app/profile/ai2.bsky.social/post/3mseffby6so23📸 140+ people joined us during #SeattleTechWeek to learn how AI gets built at Ai2. Research lead Iz Beltagy walked through what it takes to train a truly open LLM—from large-scale training runs to eval before release. Thanks to everyone who came + asked thoughtful questions. 🤝05 Aug 2026 20:44 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mseffby6so23https://bsky.app/profile/ai2.bsky.social/post/3mry26yzbxz2eWhen a model writes, where do its words come from? Are they new, or do they match exactly with language it saw in training? An AI-writing detector can't tell you. @tuhinchakr.bsky.social's group at Stony Brook has been dissecting AI-generated prose with our infini-gram engine. 🧵31 Jul 2026 22:52 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mry26yzbxz2ehttps://bsky.app/profile/ai2.bsky.social/post/3mrpvmg2rro2rThe organizations best positioned to use Earth-observation models – those working in conservation, food security, & disaster response – often can't run these models at the scale they need. That's an infrastructure problem. So we built the OlmoEarth Platform to solve it. 🧵28 Jul 2026 17:08 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mrpvmg2rro2rhttps://bsky.app/profile/ai2.bsky.social/post/3mrfvgbyers2kAs a nonprofit research institute dedicated to advancing open science, we're encouraged to see growing support for open models across the AI ecosystem. We believe the evidence behind advanced AI systems shouldn’t be locked up in a few hands.24 Jul 2026 17:38 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mrfvgbyers2khttps://bsky.app/profile/ai2.bsky.social/post/3mrfvf5isi22kAs a nonprofit research institute dedicated to advancing open science, we're encouraged to see growing support for open models across the AI ecosystem. We believe the evidence behind advanced AI systems shouldn’t be locked up in a few hands.24 Jul 2026 17:38 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mrfvf5isi22khttps://bsky.app/profile/ai2.bsky.social/post/3mrfvckpg7c2cAs a nonprofit research institute dedicated to advancing open science, we're encouraged to see growing support for open models across the AI ecosystem. We believe the evidence behind advanced AI systems shouldn’t be locked up in a few hands.24 Jul 2026 17:36 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mrfvckpg7c2chttps://bsky.app/profile/ai2.bsky.social/post/3mrfvaaik5k2hAs a nonprofit research institute dedicated to advancing open science, we're encouraged to see growing support for open models across the AI ecosystem. We believe the evidence behind advanced AI systems shouldn’t be locked up in a few hands.24 Jul 2026 17:35 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mrfvaaik5k2hhttps://bsky.app/profile/ai2.bsky.social/post/3mrdbvzzw532uCurious how AI gets built at Ai2? For #SeattleTechWeek, we're opening our Northlake office for a panel with the researchers building our open models. They'll talk through the deep technical work behind them, with time after to chat about what you're working on & network. 👇23 Jul 2026 16:44 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mrdbvzzw532uhttps://bsky.app/profile/ai2.bsky.social/post/3mr6ctywwrw2qTwo updates to Asta, our ecosystem of AI agents for science: a one-click handoff from AutoDiscovery to Asta’s data analysis tools, & paper search that evaluates its own results + searches again when they fall short. 🧵21 Jul 2026 17:17 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mr6ctywwrw2qhttps://bsky.app/profile/ai2.bsky.social/post/3mqud2rxp242kWhat does it actually take to build cutting-edge AI systems? On July 30 during #SeattleTechWeek, the researchers behind Ai2's open models sit down to talk through the deep technical work behind them. 🧵17 Jul 2026 17:55 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mqud2rxp242khttps://bsky.app/profile/ai2.bsky.social/post/3mqs772jbje2sThis week we hosted researchers from @cgiar.org, @servirglobal.bsky.social, NASA Harvest, & @msftresearch.bsky.social at our office to explore OlmoEarth for food security and natural resource management challenges. The CGIAR teams brought their own use cases and put OlmoEarth to work. 🧵16 Jul 2026 21:40 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mqs772jbje2shttps://bsky.app/profile/ai2.bsky.social/post/3mqpy4fabo72tIt was an honor to bring OlmoEarth to the Nature Positive Summit (@npinitiative.bsky.social) this week in Kumamoto, Japan, with our partners The Group on Earth Observations (@earthobservations.bsky.social). 🧵16 Jul 2026 00:28 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mqpy4fabo72thttps://bsky.app/profile/ai2.bsky.social/post/3mqp7gxwg6l2iWe built SciArena to test how well AI models handle scientific literature questions, as judged by researchers. It's retiring July 15, and the results are in: ~1,700 users cast ~3,900 votes. Here's what they told us. 🧵15 Jul 2026 17:06 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mqp7gxwg6l2ihttps://bsky.app/profile/ai2.bsky.social/post/3mqkee22ilt2qOur @skylightmarine.bsky.social team is rolling out Shippy, an AI agent for the people protecting our ocean in real time. A wrong answer at these stakes can send a patrol vessel miles off course. Here's the architecture we built so analysts can trust it. 🧵13 Jul 2026 18:51 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mqkee22ilt2qhttps://bsky.app/profile/ai2.bsky.social/post/3mqctwrpjfb2qolmOCR 2 is now in the Ai2 Playground—our home for our fully open text, video, and image understanding models. 🧵10 Jul 2026 19:09 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mqctwrpjfb2qhttps://bsky.app/profile/ai2.bsky.social/post/3mqciujmltr2nOlmoEarth turns Earth data into insights—in hours, not years. But it's only as powerful as the people using it. So, we brought them into one room. 🧵10 Jul 2026 15:50 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mqciujmltr2nhttps://bsky.app/profile/ai2.bsky.social/post/3mqafwtnc7f2h"Embeddings only get you so far… if you want the next level in performance, I think fine-tuning is the way to go." Ai2 research scientist @pjreddie.bsky.social explains how our partners customize OlmoEarth – our open-source Earth-observation models – to map crops, wildfire risk, & more. 👇09 Jul 2026 19:53 +0000at://did:plc:i4kytxgsu3yfsrt2ml3o7tgq/app.bsky.feed.post/3mqafwtnc7f2h