Qwen3 235B A22B Instruct 2507
Qwen · released Jul 21, 2025 · Qwen/Qwen3-235B-A22B-Instruct-2507
- Type
- Open weightsApache License 2.0
- Params
- 235B
- Context
- 262K
22B active per word · about 197K words of context
Our take
Written Sep 2, 2026Qwen3 is a large mixture-of-experts model from Alibaba with 235.1 billion total parameters and 22 billion active per token, released in 2025 under an Apache licence. It scores strongly on competitive programming and arena coding tests, and is available from 16 hosted providers.
Pick this for frontier coding tasks, long-context work up to 262,144 tokens, or cost-sensitive hosted deployment where an open licence matters. Use it for self-hosting when Apache 2.0 flexibility is required. Skip it if you need image, video or audio input, or if creative writing quality is your main priority — it lags there by a wide margin against its own coding scores.
The case for it
- 80.4% pass rate on contamination-free competitive programming problems.
- Arena scores above 1415 across coding, hard prompts, maths and instruction following, all from late 2026 evaluations.
- 235.1 billion total parameters with 22 billion active per token — massive scale with selective activation.
- Apache 2.0 licence and 16 current offers across 7 providers.
The case against it
- Creative writing arena score is 92.67 points below its coding peak — a clear gap.
- Text-only; no image, video or audio input.
- Throughput and pricing vary sharply by provider: no single host dominates both speed and cost.
How good is it?
EverydayGeneral questions and everyday reasoning
Arena Text (overall)77th of 168 · 1422
Also on this board: 1403 (Sep 25, 2026). Read the pair, not the higher one.
CodingWriting and fixing code on its own
Arena Coding74th of 168 · 1472
Also on this board: 1445 (Sep 25, 2026). Read the pair, not the higher one.
AgenticPlanning, calling tools, staying on task
Not yet scored on Arena Agent.
WritingDrafting and rewriting prose
Arena Creative Writing86th of 168 · 1378
Arena Creative Writing is the only board that has scored it for this.
Also on this board: 1366 (Sep 25, 2026). Read the pair, not the higher one.
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 model7 scoresEvery figure we hold, from 7 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
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Apple M1 Pro (16-core GPU) · 32 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Comfortable fit
Apple M3 Ultra (80-core GPU) · 512 GB
Room to spare. 229.7 GB spare means a 10% error in the size would not 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 between 1 hour and 5 days ago — each listing carries its own date.
Google Vertex AI, through OpenRouter
The only listing at 262K of context — the other 9 in the table below are not like-for-like. 8 cheaper rows there are outside that comparison: a different context length or a different quantisation.
- per 1M tokens
- $0.22 in / $0.88 out
- Context served
- 262K
- Throughput
- ~23 tok/s
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| GMICloudfp8Through OpenRouter | $0.087 / $0.35checked 1 hour ago | 262K236K max reply | 31 tok/s | No | Yesunknown period | Unknown |
| DeepInfrafp8Direct and through OpenRouter | $0.090 / $0.55checked 1 hour ago | 262K16K max reply through OpenRouter | 10 tok/sthrough OpenRouter | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterConfirmed |
| Novita AIfp8Direct and through OpenRouter | $0.090 / $0.58checked 1 hour ago | 131K16K max reply through OpenRouter | 21 tok/sthrough OpenRouter | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterConfirmed |
| Alibaba CloudThrough OpenRouter | $0.15 / $0.60checked 1 hour ago | 131K33K max reply | 35 tok/s | No | Yesunknown period | Unknown |
| Nebius AI Studiofp8Through OpenRouter | $0.20 / $0.60checked 43 hours ago | 262K236K max reply | 39 tok/s | No | No | Confirmed |
| Venice AIfp8Through OpenRouter | $0.15 / $0.75checked 1 hour ago | 128K16K max reply | 11 tok/s | No | No | Confirmed |
| Parasailfp8Through OpenRouter | $0.14 / $0.80checked 1 hour ago | 131K118K max reply | 32 tok/s | No | No | Confirmed |
| StreamLakeThrough OpenRouter | $0.21 / $0.84checked 1 hour ago | 128K32K max reply | 29 tok/s | No | Yesunknown period | Unknown |
| Google Vertex AIus-south1Through OpenRouter | $0.22 / $0.88checked 1 hour ago | 262K16K max reply | 23 tok/s | No | No | Confirmed |
| OpenRouterOpenRouter's own listing | $0.46 / $1.82checked 5 days ago | 131K | not measured | Unknown | Unknown | Unknown |
DeepInfra
Novita AI
Nebius AI Studio
OpenRouterAcross the 10 listings we hold: 9 say they do not train on prompts (2 of them only through OpenRouter), 0 say they do and 1 does not say. 6 appear in the zero-retention registry we check (2 of them only through OpenRouter); 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 |
|---|---|---|---|
| GMICloudfp8Through OpenRouter | ✓ | ✓ | ✓ |
| DeepInfrafp8Direct and through OpenRouter | ✓ | ✓ | ✓ |
| Novita AIfp8Direct and through OpenRouter | ✓ | ✓ | ✓ |
| Alibaba CloudThrough OpenRouter | ✓ | ✓ | ✗ |
| Nebius AI Studiofp8Through OpenRouter | ✓ | ✓ | ✓ |
| Venice AIfp8Through OpenRouter | ✓ | ✓ | ✓ |
| Parasailfp8Through OpenRouter | ✓ | ✓ | ✓ |
| StreamLakeThrough OpenRouter | ✓ | ✓ | ✓ |
| Google Vertex AIus-south1Through OpenRouter | ✓ | ✓ | ✓ |
| OpenRouterOpenRouter's own listing | ✓ | ✓ | ✗ |
Tool calling: 10 of 10 listings say yes. JSON output: 10 of 10 listings say yes. Strict schema: 8 of 10 listings say yes, 2 say no.
Models people weigh against Qwen3 235B A22B Instruct 2507
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 10 listings does not say whether it trains on prompts, and 2 answer only through OpenRouter, not for their own listing.
- 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
- Qwen/Qwen3-235B-A22B-Instruct-2507
- Architecture
- Mixture of experts
- Takes in, gives back
- Text in, text out
- Catalogue slug
- qwen-qwen3-235b-a22b-instruct-2507