gpt-oss-120B
Model ● Covered in 9 stories + Follow
gpt-oss-120B is an open-weight reasoning language model with 120 billion parameters, released under the Apache 2.0 license. Recent coverage describes it being deployed and benchmarked in multiple contexts, including instruction-following evaluations (with more than 75% failure to follow reasoning-step instructions), IT automation diagnostics where it exhibited higher averaged failure-mode counts on failed traces, and agent memory studies where it showed improved task completion with curated retrieval guidelines at lower token overhead than full guideline injection. The model has also been referenced alongside fine-tuning and optimization workflows via Together AI for domain specialization and local efficiency comparisons.
Updated 10 September 2026
Specifications
No specifications recorded yet.
Latest developments
Q3 2026
- How Much Memory Does Your Agent Actually Need?
- webAI Releases TwIL-LM: A 1.7B and 3B Formal-Logic Model Family for Autoformalization on Local Hardware
Q1 2026
Q4 2025
Q3 2025
OpenAI releases two open-weight models gpt-oss-120b and gpt-oss-20b under Apache 2.0 license Open source release
OpenAI Releases GPT-OSS Open-Weight Reasoning Models Under Apache 2.0 License Open source release
Relationships
Products & technology
- OpenAI develops this model · 5 sources
- Together AI deploys this model · 1 source
- Together AI integrated with this model · 1 source