Olmo 3.1 32B Instruct
This is the 32B instruction-following release in Ai2's Olmo 3.1 line.
What the release provides
It uses Dolma 3 pretraining and Dolci post-training data. The final checkpoint follows supervised training, preference optimization and reinforcement learning with verifiable rewards. The weights use Apache 2.0.Olmo 3.1 32B Instruct official model card
How to use it well
Practical guidance. Use this checkpoint when the application needs direct instruction-following responses. Keep the prompt template and model revision fixed when comparing it with the Think version.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | allenai/Olmo-3.1-32B-Instruct |