Ai2

It was an incredible panel at #NVIDIAGTC yesterday! Here’s @hannahajishirzi sharing the stage with Jensen Huang, explaining how our open model flow enables infinite customization.

The future of AI isn't being built in a vacuum. 🤝 NVIDIA Founder and CEO Jensen Huang sat down with the builders from AMP PBC, @bfl_ml, @Cursor_ai, @LangChain, @MistralAI, @EvidenceOpen, @Perplexity_ai, @Reflection_AI, @Thinkymachines & @Allen_ai to discuss the rapid rise of open frontier models. 🔗 Learn more: #open-models-panel" target="_blank" rel="nofollow noopener">blogs.nvidia.com/blog/gtc-2026-…

Plus, we’re releasing a VLA leaderboard with 657 results across 17 benchmarks and 509+ model configurations from 1,704 papers. The entire framework – evaluation harness, Docker configs, and leaderboard – is open source under Apache 2.0. 🔗 Code: github.com/allenai/vla-ev… 🏆 Leaderboard: allenai.github.io/vla-evaluation… 📝 Paper: arxiv.org/abs/2603.13966

We also conducted a reproducibility audit of a published VLA model across 3 benchmarks—closely matching reported results while uncovering hidden requirements (termination rules, normalization stats) that can silently skew outcomes.

📢 Introducing vla-evaluation-harness—a unified, fully open framework to evaluate any VLA model on any robot simulation benchmark. Integrate your model once. Integrate the benchmark once. The full cross-evaluation matrix fills itself. 🧵

To learn more, find our paper here: huggingface.co/datasets/allen…

VLMs already have visual tokens. Letting them point by selecting those tokens turns out to be simpler, faster, & better. 🤖 Models: huggingface.co/collections/al… 📦 Data: huggingface.co/collections/al… 💻 Code: github.com/allenai/molmo2 📖 Blog: allenai.org/blog/molmopoint

Grounding lets vision-language models do more than describe—they can point to where a robot should grasp, which button to click, or which object to track across video frames. Today we're releasing MolmoPoint, a better way for models to point. 🧵

VLMs today—including our own Molmo—point via raw text strings (e.g. ""). What if pointing meant directly selecting the visual tokens instead? 🤔 Introducing MolmoPoint: Better Pointing for VLMs with Grounding Tokens 🎯 🔓models, code, data, demo all OPEN 🧵👇 Paper: allenai.org/papers/molmopo…