Olmo 3 32B Think
Allen Institute for AI · released Nov 19, 2025 · allenai/Olmo-3-32B-Think
- Type
- Open weightsApache License 2.0
- Params
- 32.2B
- Context
- 66K
about 49K words of context
Our take
Written Sep 30, 2026Olmo 3 32B Think is a downloadable text model from the Allen Institute for AI, released under terms that allow commercial use, changes and redistribution. Its placings sit in the lower part of the field on every board we hold, so treat it as something to trial on work you can judge yourself rather than a model the evidence recommends.
Reach for it when you want a model you can download and run yourself under a licence that allows commercial use, changes and redistribution, and you are willing to judge the output yourself. Its request capacity takes a long report or a stack of documents without splitting them up first, though reliable recall across all of it is unverified in our data. Skip it if you need a model that places near the top of a preference board, or if you need measured coding, maths or instruction-following quality beyond a human-preference placing.
The case for it
- You can download it and run it yourself, and the Apache License 2.0 allows commercial use, changes and redistribution, so building a product on it is permitted by the licence.
- A request capacity of 65536 tokens leaves room for a long report or a stack of documents beside the question, though reliable recall across all of it is unverified in our data.
- One hosted offer is listed, at OpenRouter, so you can try it through a host and download it later if it earns a place.
The case against it
- Its placings sit in the lower part of the field on every board we hold: 144th of 168 on Arena Text (overall) and 139th of 168 on Arena Coding, both as of 25 Sep 2026.
- Every score supplied is a human-preference placing, so a task that needs a right answer rather than a liked one needs a trial on work you can check yourself.
- At 32.2 billion parameters, all of them used on every token, this is not a model to assume will run on modest hardware — a size is not a fit verdict.
How good is it?
An open text model for reasoning and multi-step tasks, though it trails most models on everyday questions, drafting and coding.
- getting answers to everyday questionsArena Text (overall) · 144th of 168
- drafts, rewrites and editingArena Creative Writing · 145th of 168
- writing and completing codeArena Coding · 139th of 168
EverydayGeneral questions and everyday reasoning
Arena Text (overall)144th of 168 · 1306
CodingWriting and fixing code on its own
Arena Coding139th of 168 · 1360
Arena Coding is the only board that has scored it for this.
AgenticPlanning, calling tools, staying on task
Not yet scored on Arena Agent.
WritingDrafting and rewriting prose
Arena Creative Writing145th of 168 · 1265
Arena Creative Writing is the only board that has scored it for this.
These tests check whether a model follows instructions — a precondition for all the work above, but not a measure of how well that work is done.
Every published score for this model6 scoresEvery figure we hold, from 6 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Can you run it yourself?
GeForce RTX 4090 · 24 GB
Borderline fit on an estimated size. It leaves 0.2 GB spare on a size we calculated rather than measured, and a 10% error either way would change the answer.
Comfortable fit
GeForce RTX 5090 · 32 GB
Room to spare. 8.2 GB spare means a 10% error in the size would not change the answer.
Apple M1 Pro (16-core GPU) · 32 GB
Borderline fit on an estimated size. It leaves 1.4 GB spare on a size we calculated rather than measured, and a 10% error either way would change the answer.
Memory use by level
Against a 24 GB card.
What is quantisation? →This model on every device we track71 devicesThe Q4 build most people download, on each device: what the weights come to, how much context the memory leaves, and whether it runs. Smallest device that runs it first.
Check against your own machine → · Where to rent it hosted →
Or rent it from someone else
Prices checked 34 days ago
- per 1M tokens
- $0.15 in / $0.50 out
- Context served
- 66K
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| OpenRouterOpenRouter's own listing | $0.15 / $0.50checked 34 days ago | 66K | not measured | Unknown | Unknown | Unknown |
Across the 1 listings we hold: 0 say they do not train on prompts, 0 say they do and 1 does not say. 0 appear in the zero-retention registry we check; the rest are unknown to us.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| OpenRouterOpenRouter's own listing | ✗ | ✓ | ✓ |
Tool calling: 0 of 1 listings say yes, 1 says no. JSON output: 1 of 1 listings says yes. Strict schema: 1 of 1 listings says yes.
When we formed this view
Recent changes
What moved
first indexed by our pipelineEach date is the day we first saw the change, or the day the maker announced it.
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 1 of 1 listings does not say whether it trains on prompts.
- We hold no cached-input rate for any of its listings.
- We hold no batch or off-peak rate for any of its listings.
- We hold a decode speed for it, but no prompt-processing (prefill) figure, so how long the input side of a job takes is unknown to us.
Licence and identifiers
What the licence allowsApache License 2.0, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
Apache License 2.0
Fully permissive: commercial use, redistribution, and derivatives allowed. Requires attribution and a copy of the license. Includes an express patent grant.
Identifiers
- Hugging Face
- allenai/Olmo-3-32B-Think
- Architecture
- Dense
- Takes in, gives back
- Text in, text out
- Catalogue slug
- allenai-olmo-3-32b-think