Qwen3.5-35B-A3B
Qwen · released Feb 24, 2026 · Qwen/Qwen3.5-35B-A3B
- Type
- Open weightsApache License 2.0
- Params
- 36B
- Context
- 262K
3B active per word · about 197K words of context
Our take
The case for it
- About 3 billion of its 36 billion parameters work per token, so memory in use is closer to a small model than to a mid-size one — realistic on one machine.
- The licence allows commercial use, changes and redistribution (Apache License 2.0).
- Text, images and video go into the same request, so a screenshot or a clip does not have to be described in words first.
The case against it
- Lower half on the preference boards: 101st of 168 on Arena Text (overall) as of 25 Sep 2026 and 104th of 168 on Arena Coding as of 25 Sep 2026 — human-preference boards, not correctness tests.
- The cheapest listed offer is not the only consideration: rates differ between the listed hosts, so price alone cannot pick one for you.
How good is it?
EverydayGeneral questions and everyday reasoning
Arena Text (overall)101st of 168 · 1394
CodingWriting and fixing code on its own
Arena Coding104th of 168 · 1435
AgenticPlanning, calling tools, staying on task
Not yet scored on Arena Agent.
WritingDrafting and rewriting prose
Arena Creative Writing105th of 168 · 1342
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 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
Loads, but do not expect an assistant. Some of the weights sit in ordinary system memory, which is far slower than the card.22.7 GB of weights, plus 2.1 GB for the software that runs it and the smallest conversation it can hold, comes to 24.8 GB against the 22.8 GB this 24 GB device leaves free.
Comfortable fit
GeForce RTX 5090 · 32 GB
Room to spare. 6.1 GB spare means a 10% error in the size would not change the answer.
Apple M3 Pro (18-core GPU) · 36 GB
Borderline fit on an estimated size. It leaves 2.3 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 2 hours ago — each listing carries its own date.
Alibaba Cloud, through OpenRouter
Cheapest of the 3 listings we can compare like for like — at 262K of context, out of 9 in the table below. 4 cheaper rows there are outside that comparison: a different quantisation or a different context length.
- per 1M tokens
- $0.16 in / $1.30 out
- Context served
- 262K
- Throughput
- ~117 tok/s
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| Darkbloomfp4Through OpenRouter | $0.080 / $0.75checked 2 hours ago | 262K33K max reply | 54 tok/s | No | Yesunknown period | Unknown |
| Parasailfp8Through OpenRouter | $0.15 / $1.00checked 2 hours ago | 262K236K max reply | 82 tok/s | No | No | Confirmed |
| DeepInfrafp8Direct and through OpenRouter | $0.14 / $1.00checked 2 hours ago | 262K82K max reply through OpenRouter | 49 tok/sthrough OpenRouter | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterConfirmed |
| Venice AIThrough OpenRouter | $0.31 / $1.25checked 2 hours ago | 256K16K max reply | 40 tok/s | No | No | Confirmed |
| OpenRouterOpenRouter's own listing | $0.16 / $1.30checked 2 hours ago | 262K | not measured | Unknown | Unknown | Unknown |
| Alibaba CloudThrough OpenRouter | $0.16 / $1.30checked 2 hours ago | 262K66K max reply | 117 tok/s | No | Yesunknown period | Unknown |
| AtlasCloudfp8Through OpenRouter | $0.23 / $1.80checked 2 hours ago | 262K66K max reply | 26 tok/s | No | Yesunknown period | Unknown |
| SiliconFlowfp8Through OpenRouter | $0.24 / $1.80checked 2 hours ago | 262K236K max reply | 19 tok/s | No | No | Confirmed |
| Novita AIDirect | $0.25 / $2.00checked 2 hours ago | 262K | not measured | Unknown | Unknown | Unknown |
DeepInfra
OpenRouter
Novita AIAcross the 9 listings we hold: 7 say they do not train on prompts (1 of them only through OpenRouter), 0 say they do and 2 do not say. 4 appear in the zero-retention registry we check (1 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 |
|---|---|---|---|
| Darkbloomfp4Through OpenRouter | ✓ | ✓ | ✓ |
| Parasailfp8Through OpenRouter | ✓ | ✓ | ✓ |
| DeepInfrafp8Direct and through OpenRouter | ✗ | ✓ | ✓ |
| Venice AIThrough OpenRouter | ✓ | ✓ | ✓ |
| OpenRouterOpenRouter's own listing | ✓ | ✓ | ✓ |
| Alibaba CloudThrough OpenRouter | ✓ | ✓ | ✓ |
| AtlasCloudfp8Through OpenRouter | ✓ | ✓ | ✓ |
| SiliconFlowfp8Through OpenRouter | ✗ | ✓ | ✓ |
| Novita AIDirect |
Tool calling: 6 of 9 listings say yes, 2 say no, 1 publishes no parameter list. JSON output: 8 of 9 listings say yes, 1 publishes no parameter list. Strict schema: 8 of 9 listings say yes, 1 publishes no parameter list.
Models people weigh against Qwen3.5-35B-A3B
When we formed this view
Recent changes
What moved
input $0.14 → $0.15, output $1 → $1 per 1M tokensWhat 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.
- 1 of 9 listings publishes no parameter list, so what its API accepts is unknown to us.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 2 of 9 listings do not say whether they train on prompts, and 1 answers only through OpenRouter, not for its 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/Qwen3.5-35B-A3B
- Architecture
- Mixture of experts
- Takes in, gives back
- Text, images and video in, text out
- Catalogue slug
- qwen-qwen3-5-35b-a3b