We introduce Olmo 3, a new family of 7B and 32B models both Instruct and Think variants. Long chain-of-thought thinking improves reasoning tasks like math and coding.

Olmo is a series of Open language models designed to enable the science of language models. These models are pre-trained on the Dolma 3 dataset and post-trained on the Dolci datasets. We are releasing all code, checkpoints, logs (coming soon), and associated training details.

The core models released in this batch include the following:

Stage	Olmo 3 7B Think	Olmo (3/3.1) 32B Think	Olmo 3 7B Instruct	Olmo 3.1 32B Instruct
Base Model	Olmo-3-7B	Olmo-3-32B	Olmo-3-7B	Olmo-3-32B
SFT	Olmo-3-7B-Think-SFT	Olmo-3-32B-Think-SFT	Olmo-3-7B-Instruct-SFT	Olmo-3.1-32B-Instruct-SFT
DPO	Olmo-3-7B-Think-DPO	Olmo-3-32B-Think-DPO	Olmo-3-7B-Instruct-DPO	Olmo-3.1-32B-Instruct-DPO
Final Models (RLVR)	Olmo-3-7B-Think	Olmo-3-32B-Think Olmo-3.1-32B-Think	Olmo-3-7B-Instruct	Olmo-3.1-32B-Instruct

Installation

Olmo 3 is supported in transformers 4.57.0 or higher:

pip install transformers>=4.57.0

Inference

You can use OLMo with the standard HuggingFace transformers library:

python

from transformers import AutoModelForCausalLM, AutoTokenizer
olmo = AutoModelForCausalLM.from_pretrained("allenai/Olmo-3.1-32B-Think")
tokenizer = AutoTokenizer.from_pretrained("allenai/Olmo-3.1-32B-Think")
message = ["Who would win in a fight - a dinosaur or a cow named Moo Moo?"]
inputs = tokenizer(message, return_tensors='pt', return_token_type_ids=False)
# optional verifying cuda
# inputs = {k: v.to('cuda') for k,v in inputs.items()}
# olmo = olmo.to('cuda')
response = olmo.generate(**inputs, max_new_tokens=100, do_sample=True, top_k=50, top_p=0.95)
print(tokenizer.batch_decode(response, skip_special_tokens=True)[0])
>> '<think>Okay, so the question is who would win in a fight...'

For faster performance, you can quantize the model using the following method:

python

AutoModelForCausalLM.from_pretrained("allenai/Olmo-3.1-32B-Think", 
    torch_dtype=torch.float16, 
    load_in_8bit=True)  # Requires bitsandbytes

The quantized model is more sensitive to data types and CUDA operations. To avoid potential issues, it's recommended to pass the inputs directly to CUDA using:

inputs.input_ids.to('cuda')

We have released checkpoints for these models. For post-training, the naming convention is step_XXX.

To load a specific model revision with HuggingFace, simply add the argument revision:

olmo = AutoModelForCausalLM.from_pretrained("allenai/Olmo-3.1-32B-Think", revision="step_300")

Or, you can access all the revisions for the models via the following code snippet:

python

from huggingface_hub import list_repo_refs
out = list_repo_refs("allenai/Olmo-3.1-32B-Think")
branches = [b.name for b in out.branches]

Chat template

Default System Message

The default system prompt for this model is:

<|im_start|>system
You are a helpful AI assistant.<|im_end|>

Chat Format

The chat template for this model is formatted as:

<|im_start|>system
You are a helpful AI assistant.<|im_end|>
<|im_start|>user
Who would win in a fight - a dinosaur or a cow named Moo Moo?<|im_end|>
<|im_start|>assistant
<think>Okay, so the question is who would win in a fight between a dinosaur and a cow named Moo Moo.
Hmm, first I need to break this down. Let me think about the different factors involved here..... </think>
Moo Moo the cow would certinaly win.<|im_end|>
<|endoftext|>

Model Description

Developed by: Allen Institute for AI (Ai2)
Model type: a Transformer style autoregressive language model.
Language(s) (NLP): English
License: This model is licensed under Apache 2.0. It is intended for research and educational use in accordance with Ai2's Responsible Use Guidelines.
Contact: Technical inquiries: olmo@allenai.org. Press: press@allenai.org
Date cutoff: Dec. 2024.

Model Sources

Project Page: https://allenai.org/olmo
Repositories:
- Open-Instruct for DPO and RLVR: https://github.com/allenai/open-instruct
- OLMo-Core for pre-training and SFT: https://github.com/allenai/OLMo-core
- OLMo-Eval for evaluation: https://github.com/allenai/OLMo-Eval
Paper: [TBD]

Evaluation

Benchmark	Olmo 3.1 32B Think	Olmo 3 Think 32B SFT	Olmo 3 Think 32B DPO	Olmo 3 Think 32B	Qwen 3 32B	Qwen 3 VL 32B Thinking	Qwen 2.5 32B	Gemma 3 27B Instruct	Gemma 2 27B Instruct	Olmo 2 32B Instruct	DeepSeek-R1-Distill-Qwen-32B
Math
MATH	96.2	95.6	95.9	96.1	95.4	96.7	80.2	87.4	51.5	49.2	92.6
AIME 2024	80.6	73.5	76.0	76.8	80.8	86.3	15.7	28.9	4.7	4.6	70.3
AIME 2025	78.1	66.2	70.7	72.5	70.9	78.8	13.4	22.9	0.9	0.9	56.3
OMEGA	53.4	43.1	45.2	50.8	47.7	50.8	19.2	24.0	9.1	9.8	38.9
Reasoning
BigBenchHard	88.6	88.8	89.1	89.8	90.6	91.1	80.9	82.4	66.0	65.6	89.7
ZebraLogic	80.1	70.5	74.5	76.0	88.3	96.1	24.1	24.8	17.2	13.3	69.4
AGI Eval English	89.2	85.9	87.8	88.2	90.0	92.2	78.9	76.9	70.9	68.4	88.1
Coding
HumanEvalPlus	91.5	90.0	91.6	91.4	91.2	90.6	82.6	79.2	67.5	44.4	92.3
MBPP+	68.3	66.7	67.2	68.0	70.6	66.2	66.6	65.7	61.2	49.0	70.1
LiveCodeBench v3	83.3	75.8	81.9	83.5	90.2	84.8	49.9	39.0	28.7	10.6	79.5
IF
IFEval	93.8	83.9	80.6	89.0	86.5	85.5	81.9	85.4	62.1	85.8	78.7
IFBench	68.1	37.0	34.4	47.6	37.3	55.1	36.7	31.3	27.8	36.4	23.8
Knowledge & QA
MMLU	86.4	85.3	85.2	85.4	88.8	90.1	84.6	74.6	76.1	77.1	88.0
PopQA	30.9	33.1	37.0	31.9	30.7	32.2	28.0	30.2	30.4	37.2	26.7
GPQA	57.5	55.7	57.6	58.1	67.3	67.4	44.6	45.0	39.9	36.4	61.8
Chat
AlpacaEval 2 LC	69.1	69.1	78.6	74.2	75.6	80.9	81.9	65.5	39.8	38.0	26.2
Safety	83.6	64.8	65.3	68.8	69.0	82.7	81.9	68.6	74.3	83.8	63.6

Model Details

Stage 1: SFT

supervised fine-tuning on the Dolci-Think-SFT-7B dataset. This dataset consits of math, code, chat, and general knowledge queries.
Datasets: Dolci-Think-SFT-7B, Dolci-Instruct-SFT-7B

Stage 2:DPO

direct preference optimization on the Dolci-Think-DPO-7B dataset. This dataset consits of math, code, chat, and general knowledge queries.
Datasets: Dolci-Think-DPO-7B, Dolci-Instruct-DPO-7B

Stage 3: RLVR

reinforcement learning from verifiable rewards on the Dolci-Think-RL-7B dataset. This dataset consits of math, code, instruction-following, and general chat queries.
Datasets: Dolci-Think-RL-7B, Dolci-Instruct-RL-7B

Inference & Recommended Settings

We evaluated our models on the following settings. We also recommend using them for generation:

temperature: 0.6
top_p: 0.95
max_tokens: 32768

transformers Example

python

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "allenai/Olmo-3.1-32B-Think"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    device_map="auto",
)

prompt = "Who would win in a fight - a dinosaur or a cow named MooMoo?"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

outputs = model.generate(
    **inputs,
    temperature=0.6,
    top_p=0.95,
    max_new_tokens=32768,
)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))

vllm Example

python

from vllm import LLM, SamplingParams

model_id = "allenai/Olmo-3.1-32B-Think"
llm = LLM(model=model_id)

sampling_params = SamplingParams(
    temperature=0.6,
    top_p=0.95,
    max_tokens=32768,
)

prompt = "Who would win in a fight - a dinosaur or a cow named MooMoo?"
outputs = llm.generate(prompt, sampling_params)
print(outputs[0].outputs[0].text)

Bias, Risks, and Limitations

Like any base language model or fine-tuned model without safety filtering, these models can easily be prompted by users to generate harmful and sensitive content. Such content may also be produced unintentionally, especially in cases involving bias, so we recommend that users consider the risks when applying this technology. Additionally, many statements from OLMo or any LLM are often inaccurate, so facts should be verified.

Citation

@misc{olmo2025olmo3,
title={Olmo 3},
author={Team Olmo and Allyson Ettinger and Amanda Bertsch and Bailey Kuehl and David Graham and David Heineman and Dirk Groeneveld and Faeze Brahman and Finbarr Timbers and Hamish Ivison and Jacob Morrison and Jake Poznanski and Kyle Lo and Luca Soldaini and Matt Jordan and Mayee Chen and Michael Noukhovitch and Nathan Lambert and Pete Walsh and Pradeep Dasigi and Robert Berry and Saumya Malik and Saurabh Shah and Scott Geng and Shane Arora and Shashank Gupta and Taira Anderson and Teng Xiao and Tyler Murray and Tyler Romero and Victoria Graf and Akari Asai and Akshita Bhagia and Alexander Wettig and Alisa Liu and Aman Rangapur and Chloe Anastasiades and Costa Huang and Dustin Schwenk and Harsh Trivedi and Ian Magnusson and Jaron Lochner and Jiacheng Liu and Lester James V. Miranda and Maarten Sap and Malia Morgan and Michael Schmitz and Michal Guerquin and Michael Wilson and Regan Huff and Ronan Le Bras and Rui Xin and Rulin Shao and Sam Skjonsberg and Shannon Zejiang Shen and Shuyue Stella Li and Tucker Wilde and Valentina Pyatkin and Will Merrill and Yapei Chang and Yuling Gu and Zhiyuan Zeng and Ashish Sabharwal and Luke Zettlemoyer and Pang Wei Koh and Ali Farhadi and Noah A. Smith and Hannaneh Hajishirzi},
year={2025},
eprint={2512.13961},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2512.13961},
}

Model Card Contact

For errors in this model card, contact olmo@allenai.org.