StartLux-Decision-v1.1-27B-Instant: Shanghai decision model tops Hugging Face Decision Index
Shanghai AI startup StartLux (原点星辉) released a 27B Qwen3.5-based decision model with typed-decision outputs that scored 69.1 on Hugging Face's Decision Index 0.3.2, reported as the new leaderboard leader.
Why it matters
A 27B open-weight decision-focused model topping a frontier index — decision models are emerging as the gating/approval layer for agentic systems, and this is a fresh entrant in that niche.
What you could build with it
A startup building agent infrastructure could ship an approval-gate microservice powered by decision models: every high-stakes agent action (refunds, deletions, sends) gets a typed yes/no judgment with a reason, giving ops teams an audit trail without a human in the loop for routine cases.
Does it hold up?
Too early to judge: the record rests on the HF Decision Index number in vendor-adjacent reporting; no independent benchmark or production deployment reported yet. Note the CC-BY-NC-4.0 licence, unlike the Apache-2.0 v1 family.
Built with StartLux-Decision-v1.1-27B-Instant
- 十余天两度刷新全球纪录！上海AI模型登顶国际决策模型榜单上观新闻 · Reports StartLux-Decision-v1.1-27B Instant scoring 69.1 on HF Decision Index 0.3.2, topping the leaderboard — the story driving today's interest.
- Jev Dethroned: StartLux China Open-Source Decision-Making Model Surges to Top Spot36kr · 36Kr English coverage of StartLux-Decision topping the Decision Index and beating Jev 1.13 35-1 in a chess match, with open weights released across 0.8B to 27B sizes.
- StartLux-Decision: AutoThink Decision Models with RL for Calibrated Deliberative Decisionsarxiv · The team's paper detailing the RL training pipeline, bidirectional attention gains, and Decision Index 0.2.1/0.3 evaluation results behind the 27B Instant model.
Learn more
First spotted on hf: source.
More AI models
OpenAI publishes 722 AI-generated math manuscripts from an unreleased modelOpenAI released 722 mathematical manuscripts in 372 result families to GitHub under Apache-2.0, produced by an…models · JEV 0.74Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73NVIDIA Isaac Lab 3.0 GA + Isaac OS open runtime + Agile ONE SN humanoidMajor Isaac platform expansion at GTC Berlin Oct 21: Isaac Lab 3.0 is generally available as the open-source robot…models · JEV 0.73StepFun Step-5-Preview — open BF16 weights releasedStepFun published the full BF16 weights for its flagship Step-5-Preview agentic model (600B-parameter sparse MoE, 27B…models · JEV 0.73NVIDIA Nemotron 3.5 Retriever collection: Embed 8B/1B, Rerank 8B, Parse 631MNVIDIA's Oct 21 collection drop for agentic retrieval: Nemotron-3.5-Embed in 8B BF16, 1B BF16 and 1B NVFP4…models · JEV 0.72Kimi K3.1 Pro: Moonshot's 2.8T open-weight flagship officially launchesMoonshot AI officially launched Kimi K3.1 Pro on Oct 23, a mixture-of-experts model with 2.8 trillion total / 104…models · JEV 0.72
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.