BAAI open-sources AREX-2, a 27B long-horizon agent model with 262K context
AdinaYakup · x · 2026-09-30
BAAI (Beijing Academy of Artificial Intelligence) has released AREX-2, a long-horizon agent model now on Hugging Face:
- 27B parameters, licensed under Apache 2.0
- Built on Qwen3.8 27B
- 262K context window
- Designed for long-horizon agent tasks, learning from execution feedback
Notable for developers needing a commercially usable, open-source agent model for local deployment.
Related event: BAAI Open-Sources AREX-2, a 27B Long-Horizon Agent Model(3 posts)→
More from coding & agent
- Incident Arena benchmark puts coding agents on call for incident response — amankhan · 2026-10-01
- Google: Gemini 4 Argon agents migrate 800k+ lines of C/C++ to Rust, free 300 TiB — xennygrimmato_ · 2026-10-01
- OpenAI's CUA leads on new agent stack: superhuman computer use, Dots cloud computers, Decisions API — OpenAIDevs · 2026-10-01
- AI coding agents use fast WASM sims as oracles to port game code — gandamu_ml · 2026-10-01
- RLTL;DR paper: self-generated feedback breaks Pass@128=0 barrier to 14-31% Pass@1 — burny_tech · 2026-10-01
- Dev claims 'Sonnet 5.5' hits 120-130tps with 100% pass rate on his end-to-end app benchmark — julianharris · 2026-10-01