Xiaohongshu open-sources Iris: 35B search agent nears trillion-param models on BrowseComp
小红书技术REDtech · wechat · 2026-09-14
Xiaohongshu's AllSpark team open-sourced Iris, a search agent scoring 82.2 on BrowseComp at just 35B parameters (near trillion-param Kimi-K2.6) and 88.6 for the 397B Iris-pro, topping open-source on BrowseComp, DeepSearchQA and HLE. Key recipe: auto-generating training questions from hyperlink structures with entity paraphrasing to block string-matching shortcuts; SFT-RL alternating training with in-cluster reward judging/summarization and resumable long trajectories; plus surprising generalization to General ToolUse and Cowork tasks. Weights, eval code and paper released.
More from coding & agent
- Ex-Stripe Exec: Ready Your Business for Agent Buyers With Checkout, Payment Element and Invoices — jeff_weinstein · 2026-09-21
- Dev clones Jev by LoRA-tuning Qwen3.5 4B on 25M synthetic tokens for $2 GPU hours — nato_nob · 2026-09-21
- Same prompt showdown: ChatGPT vs Claude vs Figma Make — Tegadesigns · 2026-09-21
- Google's Agent Substrate detailed: AX app layer on managed agentic compute infra — rakyll · 2026-09-21
- After 15 years of better dashboards, agents can now execute finance transactions themselves — LexSokolin · 2026-09-21
- Why sandbox-as-a-service startups are booming — and whether labs will just build it themselves — dejavucoder · 2026-09-21