Axion 1.0: open agent models trained with two SFT stages and RL
ByteDance-linked researchers released Axion 1.0, open agent foundation models in 35B and 122B mixture-of-experts sizes, trained with two supervised fine-tuning stages followed by reinforcement learning on about 18,000 examples, with the full data, training code and grading scripts published.
Why it matters
The headline is the recipe, not the size: about 90 GPU-hours of SFT for the 35B model, no mid-training, preference tuning or distillation, yet the authors report 59.0 on full Humanity's Last Exam for the 122B version.
What you could build with it
A small research team could rerun the Axion recipe on a niche domain corpus, for example legal filings or materials-science papers, and ship a specialist research agent that searches, reads sources and runs code to answer hard questions, with the training cost low enough to iterate weekly.
Does it hold up?
Too early to judge: the recipe and code are public and the 35B checkpoint is a community reproduction, while the 122B weights have not been released, so independent reproduction is still pending.
Built with Axion 1.0
- Axion 1.0: what the open research agent is good forarticle · Independent walkthrough of the three-stage recipe, the reported benchmarks, and the caveat that the downloadable 35B checkpoint is a community reproduction, not an official release.
- bytedance/axion1_0github · 19 ★ · Official release repo with data construction code, prompts, two-stage SFT, asynchronous RL and grading scripts.
Learn more
First spotted on github: source.
More AI models
VoiceStudio: open-source, fully-local ElevenLabs alternative for voice cloning, dubbing, and audiobooksdebpalash/VoiceStudio is a desktop app that runs voice cloning, voice design, video dubbing, dictation, transcription…models · JEV 0.75Tencent 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 0.73NVIDIA 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 0.72NVIDIA Cosmos 3.5 world models: now openly availableNVIDIA's Cosmos 3.5 family of open world foundation models for robots, autonomous vehicles and vision agents is now…models · JEV 0.71
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.