whatmodel.sh / OLMo 2 32B Instruct (0325)

OLMo 2 32B Instruct (0325)

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

OLMo 2 32B Instruct (0325) is an open-weights language model from Ai2, released March 2025. Its published size is 32.2B parameters. Licence: apache-2.0.

Catalogue identity notes

Comparison-only. Arena entry 'olmo-2-0325-32b-instruct' (text). LMArena's leaderboard table (arena-leaderboard space @6f2eeeb) labels 'olmo-2-0325-32b-instruct' as 'OLMo-2-0325-32B-Instruct' (Allen AI, Apache-2.0), linking https://huggingface.co/allenai/OLMo-2-0325-32B-Instruct. Maker repo allenai/OLMo-2-0325-32B-Instruct @b960243: licence tag apache-2.0. 32,234,279,936 parameters from the Hugging Face safetensors metadata (dense, so active = total). Released 2025-03: month the maker's Hugging Face repo was created (createdAt 2025-03-12).

49.3±1.3#356 of 719 LLMsfrom 14 benchmark results across 5 sources

$ how it compares

The LLMs ranked either side of OLMo 2 32B Instruct (0325), out of 828 with a score, then more from Ai2.

#modelscorewhereRAMreleased
425Mistral Small Instruct 22B (2409) Mistral AI49±2local~15GB2024-09
426Qwen2-Math 72B Instruct Alibaba49±1local~32GB2024-08
427OLMo 2 32B Instruct (0325) Ai249±1local~21GB2025-03
428OpenThinker2 7B Open Thoughts49±2local~6.4GB2025-04
429Aya 23 35B Cohere For AI49±3local~22GB2024-05
more from Ai2
259Olmo 3.1 32B Think Ai258±1local~21GB2025-12
262Olmo 3 32B Think Ai258±1local~21GB2025-11
314Molmo 2 8B Ai255±2local~6.8GB2025-12

see all LLMs · find the ones that fit your machine

$ what it's good at

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #269 of 393 · 27% of its score · 3 tests

Beats Ministral 3 14B Reasoning 2512, Ministral 3 14B, Gemma 4 E4B here, though they rank higher overall. Trails Flan-T5 XXL, LFM2 1.2B, Granite 4.0 H 1B here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
LMArena (text)1251.6 ✎▁▁▁▂▃▃▃▆▆▇█▃ beats 93/379avg 1268▇▇▇▇▇▇ 14%blunt
AlpacaEval 2 length-controlled lab42.8% ✎▃▃·▁▁█▇▅·▁·▁ beats 16/22medium · avg 42%▇▇▇▇ 10%at its level
IFEval prompt-level loose lab85.6% ✎▁▃▁▃▁▄▁▁█▃▃▁ beats 18/22easy · avg 80%▇ 2.8%typical
math▁▂▃▅▆▅▄█▄▁▁▁ #318 of 398 · 20% of its score · 3 tests

Beats Athene-70B here, though they rank higher overall. Trails Gemma 7B (base), DeepSeek R1 Distill Qwen 1.5B, Tulu 2 DPO 13B here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
MATH (subset unspecified) lab +149.7% ✎▃▃█▃▂▆·····▁ beats 13/22medium · avg 61%▇▇▇ 8.6%at its level · several runs
GSM8K lab87.6% ✎▂▃··▃▂▅█▆▃▂▃ beats 16/22very easy · avg 88%▇▇▇ 7.6%sharp
OMEGA (math generalisation) lab9.8% ✎▅···▃▃···▃▃█ beats 1/8hard · avg 24%▇ 3.7%typical
other tests7 not in a task · 52% of its score
testresultvs the others in its rundifficultycountswhy
BIG-Bench Hard lab70.6% ✎▁▄·▁▁·▅█·▄▄▁ beats 12/22easy · avg 75%▇▇▇▇▇▇▇▇ 20%sharp · at its level
MMLU lab77.3% ✎▁▂·▁·▁▁█·▃▃▅ beats 14/22easy · avg 77%▇▇▇▇▇▇▇ 19%at its level
PopQA lab37.5% ✎▁▁▁▃▃▃▁█▁▃·▄ beats 17/21hard · avg 32%▇▇ 3.9%blunt · at its level
Open PT-LLM Leaderboard average (9 Portuguese tasks)74.8% ✎▃▁▁▁▁▁▁▁▂█▅▁ beats 224/276medium · avg 73%▇ 3.1%at its level · noisy
TruthfulQA MC2 lab73.2% ✎▃▃▂·▃▃·▅█▂▅▃ beats 21/21medium · avg 63%▇ 3.0%blunt · at its level
EuroEval HellaSwag (English), test split, 5-shot generative MC, accuracy80.6% ✎▃·▁▁▂▁▂▅█▇▄▁ beats 74/100medium · avg 68%▇ 2.5%at its level · noisy
EuroEval Life in the UK (English knowledge MC), test split, 5-shot, accuracy90.8% ✎▁▁▁▁▁▁▁▂▄█▄▃ beats 119/146easy · avg 83%▇ 1.7%noisy

$ where to run it

llama.cpp · recommended

version

  1. Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
    macOS / Linux
    curl -LsSf https://llama.app/install.sh | sh
    Windows
    irm https://llama.app/install.ps1 | iex

    other ways to install ↗

  2. Run OLMo 2 32B Instruct (0325) · needs ~18.42 GB · from huggingface.co/allenai
    chat in the terminal
    llama cli -hf allenai/OLMo-2-0325-32B-Instruct-GGUF:Q4_K_S
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf allenai/OLMo-2-0325-32B-Instruct-GGUF:Q4_K_S

LM Studio

version

  1. Install LM Studio · Open the app once before using the lms command.

    download it (macOS / Windows / Linux) ↗

  2. Run OLMo 2 32B Instruct (0325) · needs ~18.42 GB · from huggingface.co/allenai

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/allenai/OLMo-2-0325-32B-Instruct-GGUF@Q4_K_S -y
  4. Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run OLMo 2 32B Instruct (0325) · needs ~18.42 GB · from huggingface.co/allenai
    chat in the terminal
    ollama run hf.co/allenai/OLMo-2-0325-32B-Instruct-GGUF:Q4_K_S

MLX (Mac)

  1. Install MLX (Mac) · Apple silicon Macs only, text only.
    with uv
    uv tool install mlx-lm
    or
    pip install mlx-lm
  2. Run OLMo 2 32B Instruct (0325) · needs ~18.1 GB, Mac only · from huggingface.co/mlx-community
    chat in the terminal
    mlx_lm.chat --model mlx-community/OLMo-2-0325-32B-Instruct-4bit
  3. Or start an OpenAI-compatible server at http://localhost:8080/v1
    mlx_lm.server --model mlx-community/OLMo-2-0325-32B-Instruct-4bit

Weights and documentation: huggingface.co/allenai ↗

$ how the score was worked out · 2 OLMo 2 Furious (Team OLMo,… 71%, LMArena leaderboard dataset 14%, allenai/Olmo-3.1-32B-Think mo… 7% and 2 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

sourceresultsshare of the scorelicence
2 OLMo 2 Furious (Team OLMo, Ai2) — authors' own evaluations of OLMo 2 Instruct and peer instruct models · lab-reported8▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 71%CC BY 4.0
LMArena leaderboard dataset1▇▇▇▇ 14%CC BY 4.0
allenai/Olmo-3.1-32B-Think model card · lab-reported2▇▇ 7%Apache-2.0
EuroEval leaderboards (results collection), English generative leaderboard2▇ 4%MIT
Open Portuguese LLM Leaderboard (archived)1▇ 3%Apache-2.0
prior · pull toward the middle▇ 1%
weighted average= 49.3

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json