{"id":"sensenova-roborsi-27b-preview","slug":"sensenova-roborsi-27b-preview","name":"SenseNova-RoboRSI-27B-Preview","orgId":"sensetime","family":"sensenova-roborsi-27b-preview","releaseDate":null,"availability":"announced","openWeights":false,"license":null,"paramsTotal":27000000000,"paramsSource":"slug","contextWindow":null,"knowledgeCutoff":null,"modelCardUrl":"https://x.com/SenseTime_AI/status/2112166037936312671","hfRepo":null,"modalities":["robotics","text","image"],"aliases":["SenseNova-RoboRSI-27B-Preview","roborsi-27b-preview","roborsi 27b preview","sensenova roborsi 27b","sensenova-roborsi-27b","sensenova/sensenova-roborsi-27b-preview"],"aliasNote":null,"origin":"tracked","tracked":{"trackedSince":"2026-10-21","announced":"2026-10-19","blurb":"SenseTime's 27B embodied model for robot planning and decision-making, trained from Qwen3.8-27B with experience from its open-source SenseNova-RoboRSI harness, which runs and improves robot agents in simulation and on real robots. Released alongside SenseTime's study of GPT-6 Astra on 12 simulation benchmarks and 8 real robots. Open weights announced, not yet public.","api":"none","openWeights":"announced","params":{"text":"27B (Qwen3.8-27B base)","source":"SenseNova-RoboRSI GitHub README"},"facts":[{"label":"Base model","value":"Qwen3.8-27B, trained with recursive self-improvement (RSI) on validated interaction experience"},{"label":"Harness","value":"SenseNova-RoboRSI on GitHub (Apache 2.0): shared runtime for LIBERO, LIBERO-PRO, LIBERO-Long++, Bench2Dex, DexJoCo and BEHAVIOR-1K, with code still rolling out"},{"label":"Execution","value":"Direct VLM control (System 2) or VLA-assisted (System 2 + System 1)"},{"label":"Weights","value":"Announced for Hugging Face (sensenova/sensenova-roborsi-27b-preview; README said Oct 20), not public as of Oct 21; technical report to follow"}],"claims":["Lab claim: runs at about 1/50 the cost of GPT-6 Astra.","Lab claim: 26.13 average on RoboDojo (2.84 below GPT-6 Astra), 95.0% on LIBERO-Pro with VLA-assisted execution and 96.1% with direct control, and RoboCasa365 success raised from 26.0% to 46.5% over the Qwen3.8-27B base."],"sources":[{"kind":"lab","title":"SenseTime announcement on X","url":"https://x.com/SenseTime_AI/status/2112166037936312671","date":"2026-10-19"},{"kind":"repo","title":"SenseNova-RoboRSI on GitHub","url":"https://github.com/OpenSenseNova/SenseNova-RoboRSI","date":"2026-10-21"},{"kind":"lab","title":"Has GPT-6 Astra achieved embodied intelligence? (SenseNova)","url":"https://www.sensenova.ai/has-gpt-6-astra-achieved-embodied-intelligence","date":"2026-10-19"}],"firstBenchmark":null},"organization":{"id":"sensetime","name":"SenseTime","slug":"sensetime","country":null,"url":null,"color":"#A39E8F","mark":"SE"},"price":null,"answer":{"word":"Pending benchmarks.","scope":"Announced Oct 19, 2026; no independent results yet.","lead":"As of Oct 22, 2026, SenseNova-RoboRSI-27B-Preview is pending benchmarks on isitsota: announced Oct 19, 2026, with no independent results on the boards we track yet.","confidence":"low"},"status":"PENDING_BENCHMARKS","verdicts":[]}