BREAKING: OpenAI just scored agents on the messy part of science.

This is a AI post classified by AGK Intelligence as Frontier models (a model release), kept by the AI Radar because it carries real work, not commentary.
BREAKING: OpenAI just scored agents on the messy part of science. Not paper figures. Raw gels, colony plates, microscopy, instrument exports. BioBasicsBench: 100 routine wet-lab tasks with Ligo (enzyme design). Five domains. Rubric credit, not pass/fail. At xhigh reasoning: GPT-6 Astra: 66.7% of available credit GPT-5.6 Sol: 43.8% Astra used about half the output tokens. The gap that matters is not the leaderboard. Astra caught a control plate labeled with the wrong antibiotic and rejected it. Sol treated empty colonies as a clean negative. Astra still fails the same ways product agents
Posted by Guillermo Flor (14k followers) 36 min ago · 2 likes · 147 views · view the original post on X. Kept by the AI Radar as Frontier models.
More AI work like this
- These are the best demos I've ever seen, 45 seconds of pure brilliance. Finally some… — @0xSero
- Gemini 4 Carbon was given a 3D seaplane prompt and built an infinite ocean with… — @RoundtableSpace
- Yesterday, the first details regarding Anthropic's internal models were published. — @kimmonismus
- Congratulations to Qwen3.8-27B, it beats all models including the top SOTA models from… — @TeksEdge
- OpenAI built BioBasicsBench, a test that gives AI agents the messy raw data of real lab… — @rohanpaul_ai
- GPT-6 Astra caught a wrong antibiotic label on a DNA-cloning control plate that GPT-5.6… — @choblin29
- Inkling 2 Small from @thinkymachines on ARC-AGI (Verified): — @arcprize
- 4B Nemotron 3 Nano beats Frontier models on task specific workflows 🤯 — @chadwahl
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 65.2k posts from 5k X accounts over the last 14 days, 10.5k tools, 19 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-06 20:49 UTC. Full method.