[whatmodel.sh](../../README.md) / Olmo 3.1 32B Think
# Olmo 3.1 32B Think
Ai2 · released 2025-12 · 32.2B params · apache-2.0
**score 58.2** ±0.9 · #222 of 732 · from 21 benchmark results across 5 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Olmo 3.1 32B Think among every model on that task or test, worst to best.
task vs the others rank counts tests
coding ▁▁▁▃▅▇●▇▅▄▃▁ #212 of 469 6.9% 3
chat & writing ▁▁▁▃▆▄▆●▆▇▇▃ #236 of 402 13% 3
math ▁▁▃▅█▆▄●▄▂▁▁ #114 of 431 20% 4
coding · #212 of 469 · 6.9% of its score
Beats [Qwen3.5 4B](qwen3-5-4b.md), [Falcon-H1R 7B](falcon-h1r-7b.md), [Qwen3-VL 32B Instruct](qwen3-vl-32b-instruct.md) here, though they rank higher overall. Trails [QwQ-32B-Preview](qwq-32b-preview.md), [Ministral 3 3B Reasoning 2512](ministral-3-3b-reasoning.md), [DeepSeek-Coder-V2 Instruct (0724)](deepseek-coder-v2-0724.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
MBPP+ (maker) 68.3% ▄▄····▄·▄··● beats 5/5 medium · avg 61% 2.8% at its level
LiveCodeBench v3 (maker) 83.3% ▃▃·····▃·▃·● beats 4/6 medium · avg 64% 2.3% typical
HumanEval+ (maker) 91.5% ▂·······▂·▂● beats 6/6 easy · avg 78% 1.9% sharp · noisy
chat & writing · #236 of 402 · 13% of its score
Beats [Mistral Small 4 (119B-A6.5B)](mistral-small-4.md) here, though they rank higher overall. Trails [Flan-T5 XXL](flan-t5-xxl.md), [LFM2.5 1.2B Instruct](lfm2-5-1-2b-instruct.md), [Gemma 3 4B](gemma-3-4b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LMArena (text) 1286.4 ▁▁▁▂▃▃●▆▆▇█▃ beats 114/385 avg 1260 8.0% blunt
AlpacaEval 2 length-controlled (m… 69.1% █···█·█·█·●█ beats 4/5 medium · avg 43% 4.2% sharp · at its level
IFEval (aggregation unspecified) … 93.8% ▄▄·····█▄▄·● beats 6/6 easy · avg 79% 0.5% noisy
math · #114 of 431 · 20% of its score
Beats [Claude Sonnet 4.6](claude-sonnet-4-6.md), [MiniMax-M3](minimax-m3.md), [MiMo-V2.5 (310B-A15B)](mimo-v2-5.md) here, though they rank higher overall.
test result vs the others in its run difficulty counts why
AIME 2025 (maker) 78.1% ▄··▄····█▄▄● beats 6/6 medium · avg 50% 10% sharp · at its level
AIME 2024 (maker) 80.6% ▄·····▄··▄█● beats 6/6 medium · avg 58% 7.5% sharp
OMEGA (math generalisation) (make… 53.4% ▅···▃▃···▃▃● beats 8/8 hard · avg 24% 2.0% at its level
MATH (subset unspecified) (maker) 96.2% ▅····▃▃··▃▃● beats 7/8 easy · avg 73% 1.0% noisy
other tests (11)
test result vs the others in its run difficulty counts why
Medmarks-V mean win rate (19 veri… 0.5067 ▁▄▃▃▇▄●▃▅▁▁▁ beats 18/30 avg 0.418 15% sharp
GPQA (subset unspecified) (maker) 56.7% ▃······▃·▅·● beats 5/6 hard · avg 43% 13% sharp · at its level
BIG-Bench Hard (maker) 88.6% ▄·····▄·▄·█● beats 5/6 easy · avg 77% 7.2% sharp
Medmarks-V weighted mean win rate… 50.7% ▂▃▅▂▆▃█●█▅▃▆ beats 18/34 medium · avg 41% 6.6% blunt · at its level
MMLU (maker) 86.4% ▄···▄··█··▄● beats 6/6 easy · avg 76% 5.5% typical
AGIEval English (maker) 88.8% ▄····▄·▄▄··● beats 5/5 easy · avg 76% 3.3% typical
PopQA (maker) 30.9% ▄··█···▄···● beats 4/5 hard · avg 20% 3.0% at its level
ZebraLogic (maker) 80.1% █▄····▄·▄·▄● beats 6/6 medium · avg 52% 3.0% at its level
IFBench (aggregation unspecified)… 68.1% █·██·██····● beats 5/5 medium · avg 36% 1.9% at its level
EuroEval HellaSwag (English), tes… 74.3% ▃·▁▁▂▁▂▅●▇▄▁ beats 59/100 medium · avg 67% 1.1% sharp · noisy
EuroEval Life in the UK (English … 89.9% ▁▁▁▁▁▁▁▂▄█●▃ beats 116/146 easy · avg 82% 0.6% noisy
source results share of the score
Olmo 3 (submitted / maker-reported) 14 66% ▇▇▇▇▇▇▇▇
Medmarks: A Comprehensive Open-Source L… 2 21% ▇▇▇
LMArena leaderboard dataset 1 8% ▇
allenai/Olmo-3.1-32B-Think model card (… 2 3% ▇
EuroEval leaderboards 2 2% ▇
prior: pull toward the middle 0% ▇
weighted average = 58.2
Licences: Olmo 3 ([CC BY 4.0](../../licenses/report-2512.13961v2.txt)) · Medmarks: A Comprehensive Open-Source LLM Benchmark Suite for Medical Tasks ([CC BY 4.0](../../licenses/paper-2605.01417v1-CC-BY-4.0.txt)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · allenai/Olmo-3.1-32B-Think model card ([Apache-2.0](../../licenses/allenai--Olmo-3.1-32B-Think.txt)) · EuroEval leaderboards ([MIT](https://github.com/EuroEval/leaderboards/blob/1a77f1d2f8ba1ba05ab73287fb090d76bf41b2e0/LICENSE))
## How to run it
### llama.cpp (recommended)
1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md)
```sh
# macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
# Windows
irm https://llama.app/install.ps1 | iex
```
2. Run Olmo 3.1 32B Think (Q4_K_M, needs ~19.5 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Olmo-3.1-32B-Think-GGUF/tree/87e2c446c9505e8dd96670715350fe7020075974))
```sh
llama cli -hf lmstudio-community/Olmo-3.1-32B-Think-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf lmstudio-community/Olmo-3.1-32B-Think-GGUF:Q4_K_M
```
Other versions: Q8_0 (~34.3 GB) `llama cli -hf lmstudio-community/Olmo-3.1-32B-Think-GGUF:Q8_0`
### LM Studio
1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download)
2. Run Olmo 3.1 32B Think (Q4_K_M, needs ~19.5 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Olmo-3.1-32B-Think-GGUF/tree/87e2c446c9505e8dd96670715350fe7020075974))
```sh
lms get https://huggingface.co/lmstudio-community/Olmo-3.1-32B-Think-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: Q8_0 (~34.3 GB) `lms get https://huggingface.co/lmstudio-community/Olmo-3.1-32B-Think-GGUF@Q8_0 -y` · 4bit (~18.1 GB, Mac only) `lms get https://huggingface.co/lmstudio-community/Olmo-3.1-32B-Think-MLX-4bit -y` · 8bit (~34.2 GB, Mac only) `lms get https://huggingface.co/lmstudio-community/Olmo-3.1-32B-Think-MLX-8bit -y`
### Ollama
1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download)
```sh
# Linux
curl -fsSL https://ollama.com/install.sh | sh
```
2. Run Olmo 3.1 32B Think (Q4_K_M, needs ~19.82 GB, from [ollama.com](https://ollama.com/library/olmo-3.1/tags))
```sh
ollama run olmo-3.1:32b-think
```
Other versions: Q8_0 (~39.14 GB) `ollama run olmo-3.1:32b-think-q8_0`
### MLX (Mac)
1. Install MLX (Mac). Apple silicon Macs only, text only.
```sh
# with uv
uv tool install mlx-lm
# or
pip install mlx-lm
```
2. Run Olmo 3.1 32B Think (4bit, needs ~18.1 GB, Mac only, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Olmo-3.1-32B-Think-MLX-4bit/tree/64ebeae314fcc35f69e0eec5f352444feebf1a61))
```sh
mlx_lm.chat --model lmstudio-community/Olmo-3.1-32B-Think-MLX-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model lmstudio-community/Olmo-3.1-32B-Think-MLX-4bit
```
Other versions: 8bit (~34.2 GB, Mac only) `mlx_lm.chat --model lmstudio-community/Olmo-3.1-32B-Think-MLX-8bit`
Model repository and documentation: [huggingface.co/allenai](https://huggingface.co/allenai/Olmo-3.1-32B-Think)
[interactive version](https://whatmodel.sh/model/olmo-3-1-32b-think) · [JSON](https://whatmodel.sh/api?format=json&all=1)