Olmo 3.1 32B Think

At Q4 Olmo 3.1 32B Think is estimated to need 23 GB of VRAM, at Q8 39.9 GB and at FP16 74.6 GB. These figures are estimates based on the 32.2B parameter count, covering weights plus about 15% and 0.5 GB for a short context.

Yes, a 24 GB GPU fits at Q4, estimated at 23 GB. Per estimates, a 16 GB, 12 GB or 8 GB GPU does not fit at Q4, Q8 or FP16.

Yes, with an estimated 23 GB needed at Q4, the smallest common Mac memory size that fits is 32 GB. This assumes macOS lets the GPU use about 75% of unified memory.

Yes, Olmo 3.1 32B Think has open weights, so you can self-host it for full control of your data.

Run Olmo 3.1 32B Think locally

No, Olmo 3.1 32B Think only accepts text input. It does not read images, audio, video or PDFs.

Olmo 3.1 32B Think ranks 306 on Arena Text with a score of 1286. This is Arena's own rank.

Olmo 3.1 32B Think benchmarks

Closest alternatives, matched on model kind and release date with no comparable price, are AutoGLM Phone 9B Multilingual by Z.ai, Devstral Small 2 by Mistral AI and GLM-4.6V-Flash by Z.ai.

AutoGLM Phone 9B MultilingualOlmo 3.1 32B Think vs AutoGLM Phone 9B MultilingualDevstral Small 2Olmo 3.1 32B Think vs Devstral Small 2GLM-4.6V-FlashOlmo 3.1 32B Think vs GLM-4.6V-Flash

Olmo 3.1 32B Think is open weights, released under the Apache 2.0 license.

Yes, Olmo 3.1 32B Think uses the Apache 2.0 license, a permissive license that allows commercial use.

You can download the Olmo 3.1 32B Think weights from the Hugging Face repository allenai/Olmo-3.1-32B-Think.

Run Olmo 3.1 32B Think locally

Olmo 3.1 32B Think was made by Ai2 and released on December 12, 2025.

More from Ai2

Yes, Olmo 3.1 32B Think has open weights, so you can download and run the model yourself.