{"schema":"postcutoff/event@1","as_of":"2026-11-13T13:49:00+01:00","url":"https://postcutoff.com/e/2026-05-05-ai2-molmoact-2/","md":"https://postcutoff.com/e/2026-05-05-ai2-molmoact-2/index.md","disclosure":{"written_by":"AI agents (Claude Opus 5.5 in Claude Code)","editor":"human editor","policy":"https://postcutoff.com/about/"},"license":null,"id":"2026-05-05-ai2-molmoact-2","date":"2026-05-05","date_precision":"day","short_title":"Ai2 releases MolmoAct 2, a fully open robot action-reasoning model that beats π0.5 on real-world tasks","deck":null,"takeaway":"On 2026-05-05 the Allen Institute for AI released MolmoAct 2 and MolmoAct 2-Think, open vision-language-action models built on the Molmo2-ER embodied-reasoning VLM with a flow-matching action expert, along with weights, code and 720+ hours of bimanual data.","category":"robotics","category_label":"Robotics","importance":2,"confidence":"high","status":{"key":"confirmed","labels":["Confirmed"]},"sources":[{"n":1,"title":"Ai2 blog: MolmoAct 2","url":"https://allenai.org/blog/molmoact2","type":"official","group":"primary","domain":"allenai.org"},{"n":2,"title":"arXiv 2605.02881","url":"https://arxiv.org/abs/2605.02881","type":"paper","group":"primary","domain":"arxiv.org"},{"n":3,"title":"Hugging Face: MolmoAct2 models","url":"https://huggingface.co/collections/allenai/molmoact2-models","type":"code","group":"primary","domain":"huggingface.co"},{"n":4,"title":"GitHub: allenai/molmoact2","url":"https://github.com/allenai/molmoact2","type":"code","group":"primary","domain":"github.com"},{"n":5,"title":"SiliconANGLE: Ai2 releases MolmoAct 2","url":"https://siliconangle.com/2026/05/05/ai2-releases-molmoact-2-enhancing-robot-intelligence-real-world/","type":"press","group":"press","domain":"siliconangle.com"}],"official":4,"filed":"2026-09-29","updated":"2026-09-29","orgs":["Ai2"],"title":"Ai2 releases MolmoAct 2, a fully open robot action-reasoning model that beats π0.5 on real-world tasks","summary":"On 2026-05-05 the Allen Institute for AI released MolmoAct 2 and MolmoAct 2-Think, open vision-language-action models built on the Molmo2-ER embodied-reasoning VLM with a flow-matching action expert, along with weights, code and 720+ hours of bimanual data. In Ai2's tests it reached 87.1% average success on 15 real Franka tasks (π0.5: 45.2%) and runs up to 37x faster than the original MolmoAct.","key_facts":["Paper: 'MolmoAct2: Action Reasoning Models for Real-world Deployment' (arXiv 2605.02881); weights on HF 2026-05-04/05","Real-world Franka, 15 tasks: 87.1% vs 48.4% (MolmoBot) and 45.2% (π0.5), Ai2's own evaluation","LIBERO: 97.2% (base), 98.1% (Think) vs ~86.6% for MolmoAct","Latency: ~180 ms per action call (790 ms with adaptive depth reasoning) vs 6,700 ms for MolmoAct","Molmo2-ER averages 63.8 across 13 embodied-reasoning benchmarks, ahead of GPT-5, Gemini 2.5 Pro and Gemini Robotics-ER 1.5 (Ai2)","Data: MolmoAct2-BimanualYAM (720+ h), re-annotated DROID/SO-100/BC-Z/Fractal mixture; open FAST tokenizer; code Apache-2.0"],"key_numbers":[],"tags":["vla","open-weights","open-data","robot-foundation-model","manipulation"],"science":null,"body_md":"## What happened\n\nMolmoAct 2 is the successor to Ai2's 2025 MolmoAct, which reasoned in 3D. It swaps in a stronger embodied-reasoning backbone (Molmo2-ER, trained on ~3M extra examples) and adds a separate continuous action expert, which cuts latency sharply. Everything is released: weights, training data and code, with LeRobot integration.\n\n## Why it matters\n\nIt is the most capable fully open VLA stack, with open data as well as weights. Academic labs can reproduce and extend it, unlike closed π, Gemini Robotics or Helix models. The comparisons with π0.5 are Ai2's own.","disputed":[],"related":[{"id":"2026-04-16-physical-intelligence-pi-0-7","url":"https://postcutoff.com/e/2026-04-16-physical-intelligence-pi-0-7/","date":"2026-04-16","date_precision":"day","short_title":"Physical Intelligence's π0.7 shows compositional generalization to untrained robot tasks","deck":null,"takeaway":"Emergent, promptable generalization is what would let general-purpose robots be deployed without per-task data collection.","category":"robotics","category_label":"Robotics","importance":4,"confidence":"high","status":{"key":"confirmed","labels":["Confirmed"]},"sources":3,"official":1,"filed":"2026-09-29","updated":"2026-09-29","orgs":["Physical Intelligence"]},{"id":"2026-03-16-nvidia-gtc-2026-robotics-groot-n2-cosmos-3","url":"https://postcutoff.com/e/2026-03-16-nvidia-gtc-2026-robotics-groot-n2-cosmos-3/","date":"2026-03-16","date_precision":"day","short_title":"NVIDIA previews the GR00T N2 world action model and announces Cosmos 3 at GTC 2026","deck":null,"takeaway":"NVIDIA is pitching world-model-based policies and human video as a way to trade scarce robot teleoperation data for compute.","category":"robotics","category_label":"Robotics","importance":3,"confidence":"high","status":{"key":"confirmed","labels":["Confirmed"]},"sources":10,"official":8,"filed":"2026-09-29","updated":"2026-09-29","orgs":["NVIDIA"]}],"people":[],"posts":[],"videos":[],"models":[{"id":"molmoact-2","name":"MolmoAct 2 / MolmoAct 2-Think","url":"https://postcutoff.com/m/molmoact-2/"}],"changes":[{"date":"2026-09-29","type":"filed","text":"Created"}],"provenance":{"agents":[{"model":"Claude Opus 5.5","maker":"Anthropic","tool":"Claude Code"}],"filed":"2026-09-29","run":null,"sources_read":null,"updated":"2026-09-29","human_review":null,"version":{"date":"2026-10-04"}},"gaps":[{"model_id":"gpt-6-astra","name":"GPT-6 Astra","cutoff":"2026-04","days_after":5,"in_training_data":false},{"model_id":"claude-opus-5-5","name":"Claude Opus 5.5","cutoff":"2026-06","days_after":null,"in_training_data":true},{"model_id":"gemini-4-hydrogen","name":"Gemini 4 Hydrogen","cutoff":"2026-03","days_after":35,"in_training_data":false},{"model_id":"grok-4-7","name":"Grok 4.7","cutoff":"2026-05","days_after":null,"in_training_data":true}],"short_url":null}