Prototype · closed beta
unOpen.AI
In closed beta: Roxy, a private AI assistant with permanent memory running on infrastructure the user owns. Brand, site, product, backend and conversational design.
Model families
Qwen for
multilingual systems.
Qwen's distinguishing feature is range. It ships across one of the widest ranges of sizes of any open family, which means the same architecture and tokeniser can serve a phone, a single GPU and a multi-node cluster — and that consistency is worth more in a real system than it sounds.
All Model Families →Introduction transcript. Qubi: Most real systems want more than one model size, and mixing families gets messy. This family comes in a long ladder of sizes, all sharing one design. Requests in many languages are sorted, and the easy ones step down a rung. Run on your own hardware, the weights send nothing out, and your policy decides.
Current release
Qwen today.
Alibaba Cloud ships these models as a range rather than one lead model, so this page is about the family: what it is good at, where it runs and how we evaluate it.
Last reviewed on 1 September 2026. We check the public model catalogues every week; when a new release appears, we review it against the provider's announcement and update this page.
LATEST MODELS
The newest Qwen models.
What Alibaba Cloud has shipped most recently. Each has its own page with specifications, independent ratings at every reasoning setting and listed prices.
23 Sep 2026
Qwen3.8 Max Prime
- Intelligence
- —
- Context
- 1M
- In / out per M
- $4.00 / $12.00
21 Sep 2026
Qwen3.8 Omni Flash
- Intelligence
- —
- Context
- 1M
- In / out per M
- $0.150 / $0.470
2 Sep 2026
Qwen3.8 Max (0902)
- Intelligence
- 33.6
- Context
- 1M
- In / out per M
- $2.00 / $6.00
26 Aug 2026 · open weights
Qwen3.8 Flash
- Intelligence
- —
- Context
- 1M
- In / out per M
- $0.150 / $0.470
EVERY MODEL
All 36 Qwen models, compared.
Sort by any column. Select a model for its full page.
| Qwen3.8 Max Prime | 23 Sep 2026 | — | — | 1M | $4.00 | $12.00 |
|---|---|---|---|---|---|---|
| Qwen3.8 Omni Flash | 21 Sep 2026 | — | — | 1M | $0.150 | $0.470 |
| Qwen3.8 Max (0902) | 2 Sep 2026 | 33.6 | 38 | 1M | $2.00 | $6.00 |
| Qwen3.8 Flash open | 26 Aug 2026 | — | — | 1M | $0.150 | $0.470 |
| Qwen3.8 27B open | 14 Aug 2026 | 20.1 | 39 | 1M | $0.150 | $1.88 |
| Qwen3.8 2.4T A95B open | 12 Aug 2026 | — | 38 | 1.05M | $2.00 | $6.00 |
| Qwen3.7 Flash | 27 Jul 2026 | — | — | 1M | $0.030 | $0.130 |
| Qwen3.7 Plus | 1 Jun 2026 | — | 55 | 1M | $0.320 | $1.28 |
| Qwen3.7 Max | 19 May 2026 | — | — | 1M | $1.48 | $4.43 |
| Qwen3.5 Plus 2026-04-20 | 27 Apr 2026 | — | — | 1M | $0.300 | $1.80 |
| Qwen3.6 27B open | 22 Apr 2026 | — | — | 262k | $0.320 | $3.20 |
| Qwen3.6 35B A3B open | 16 Apr 2026 | — | 137 | 262k | $0.100 | $0.900 |
| Qwen3.6 Flash | 27 Apr 2026 | — | — | 1M | $0.188 | $1.13 |
| Qwen3.6 Plus | 2 Apr 2026 | — | — | 1M | $0.325 | $1.95 |
| Qwen3.5-9B open | 2 Mar 2026 | — | 57 | 262k | $0.170 | $0.250 |
| Qwen3.5-122B-A10B open | 24 Feb 2026 | — | 134 | 262k | $0.260 | $2.08 |
| Qwen3.5-27B open | 24 Feb 2026 | — | — | 262k | $0.260 | $2.60 |
| Qwen3.5-35B-A3B open | 24 Feb 2026 | — | — | 262k | $0.140 | $1.00 |
| Qwen3.5-Flash | 25 Feb 2026 | — | — | 1M | $0.065 | $0.260 |
| Qwen3.5 397B A17B open | 16 Feb 2026 | — | 59 | 262k | $0.500 | $3.60 |
| Qwen3.5 Plus 2026-02-15 | 16 Feb 2026 | — | — | 1M | $0.260 | $1.56 |
| Qwen3 Coder Next open | 3 Feb 2026 | — | 77 | 262k | $0.120 | $0.800 |
| Qwen3 Coder Flash | 17 Sep 2025 | — | — | 1M | $0.195 | $0.975 |
| Qwen3 Next 80B A3B Instruct open | 11 Sep 2025 | — | 127 | 262k | $0.090 | $1.10 |
| Qwen3 Coder 30B A3B Instruct open | 31 Jul 2025 | — | — | 262k | $0.070 | $0.280 |
| Qwen3 30B A3B Instruct 2507 open | 29 Jul 2025 | — | — | 262k | $0.100 | $0.300 |
| Qwen3 235B A22B Thinking 2507 open | 25 Jul 2025 | — | — | 128k | $0.450 | $3.50 |
| Qwen3 Coder 480B A35B open | 23 Jul 2025 | — | — | 262k | $0.300 | $1.00 |
| Qwen3 235B A22B Instruct 2507 open | 21 Jul 2025 | — | — | 262k | $0.090 | $0.550 |
| Qwen3 14B open | 28 Apr 2025 | — | — | 41k | $0.120 | $0.240 |
| Qwen3 30B A3B open | 28 Apr 2025 | — | — | 41k | $0.120 | $0.500 |
| Qwen3 32B open | 28 Apr 2025 | — | — | 131k | $0.080 | $0.280 |
| Qwen-Plus | 1 Feb 2025 | — | — | 1M | $0.260 | $0.780 |
| Qwen2.5 Coder 32B Instruct open | 11 Nov 2024 | — | — | 33k | $0.660 | $1.00 |
| Qwen2.5 7B Instruct open | 16 Oct 2024 | — | — | 33k | $0.100 | $0.200 |
| Qwen2.5 72B Instruct open | 19 Sep 2024 | — | — | 33k | $0.360 | $0.400 |
Intelligence and coding: Artificial Analysis indices at the model's best reasoning setting, and its measured output speed. Context and prices: the OpenRouter catalogue list price in USD per million tokens. Released: the provider's date where Artificial Analysis records one, otherwise the date the catalogue first listed it. Refreshed weekly; not measured by QQuantum.ai. Select a heading to sort.
Specification
- Provider
- Alibaba Cloud, China
- Weights
- Published across the range
- Licence
- Predominantly Apache-2.0 across recent releases, with some exceptions — check the specific model card
- Sizes
- One of the widest ladders of any open family, from sub-1B up to large mixture-of-experts models
- Strengths
- Multilingual coverage including CJK, coding variants, and strong small models
- Where we use it
- Self-hosted multilingual work, and wherever a single family has to cover both an edge and a server deployment
Provider-published specifications as of 1 September 2026, not figures we measured. Model families move quickly — check the provider's own documentation before committing to a number.
Why a size ladder matters
Most production AI systems end up wanting more than one model. A small one for classification and routing, a mid-size one for the bulk of the work, a large one for the hard minority of requests. Assembling that from three different families means three tokenisers, three prompt idioms, three evaluation baselines and three sets of quirks to learn.
Qwen's range lets you build that ladder inside one family. Prompts transfer with far less rework, an evaluation harness written for one size largely holds for another, and when you move a request class from the mid-size model to the small one — which is where a great deal of cost saving actually comes from — you are changing a parameter rather than rewriting an integration.
The multilingual coverage is the second reason we reach for it. If your corpus contains Chinese, Japanese or Korean alongside European languages, this family has, in our evaluations, handled the mix better than Western-trained alternatives of comparable size, and the difference shows up immediately on a fair evaluation set rather than needing a subtle one.
The provenance question, again
Qwen is developed by Alibaba, and the assessment is structurally the same one we set out on the DeepSeek page: separate the transfer question from the behaviour question from the artefact question. Self-hosting closes the first completely — the weights are a file and they transmit nothing. The second is testable on your own material and is workload-dependent. The third is only partially answerable, for every model from every jurisdiction.
Where an organisation has a standing policy against models of Chinese origin, that policy is the answer and we build with Mistral or Llama, which occupy adjacent ground. It is a governance decision owned by people who own the risk, and it is not our place to argue anyone out of it.
Where we would choose something else
Where the deployment is a single size on a single class of hardware, the size ladder stops being an advantage and the decision comes down to capability and tooling on that one rung — which is frequently where Llama's deeper serving ecosystem wins.
Where the workload is at the hardest end of agentic or reasoning work, the leading proprietary models still lead, and a hybrid arrangement that routes that minority out is usually better than choosing one family for everything. See hybrid model routing.
INDEPENDENT RATINGS
How Qwen compares.
Third-party scores and listed prices, refreshed every week. Useful for building a shortlist; your own evaluation decides.
Composite index across their reasoning, knowledge, maths and coding evaluations · Higher is better · Linear scale from 0 to 66.0
- Gemini 4 Carbon
- 59.4
- Claude Opus 5.5
- 58.4
- GPT-6 Astra
- 56.9
- Muse Wonder Full
- 49.7
- Qwen3.8 Max
- 33.6
- Qwen3.8 27B
- 20.1
This family's highest-rated models, highlighted, against the best-rated model of each other family we cover. One composite score cannot show cost, latency, data residency or how a model handles your task.
Source: Artificial Analysis. Checked 30 October 2026. Measured independently by Artificial Analysis, not by QQuantum.ai or the providers.
USD per million output tokens · The newest releases in the family · Linear scale from 0 to $14.00
- Qwen3.8 Max Prime
- $12.00
- Qwen3.8 Omni Flash
- $0.47
- Qwen3.8 Max (0902)
- $6.00
- Qwen3.8 Flash
- $0.47
- Qwen3.8 27B
- $1.88
Input is priced separately: Qwen3.8 Max Prime $4.00, Qwen3.8 Omni Flash $0.15, Qwen3.8 Max (0902) $2.00, Qwen3.8 Flash $0.15, Qwen3.8 27B $0.15 per million tokens. Catalogue list prices; the provider's own rates, caching and batch discounts can differ.
Source: OpenRouter model catalogue. Checked 30 October 2026. Listed by OpenRouter, not measured by QQuantum.ai.
Questions
Questions about Qwen for Multilingual Work.
Is Qwen genuinely competitive?
At most sizes, yes — and its small models in particular are strong relative to their footprint, which is exactly where a lot of production value sits. As always the comparison that matters is on your task, and we run it rather than argue about it.
Can we run the small models on modest hardware?
The lower rungs of the ladder run comfortably on a single accelerator, and the smallest run on a CPU at usable speeds for batch work. That makes them a realistic option for edge deployments and for on-premise setups where a full GPU node is not justified.
How do we handle the governance question internally?
Write it down before you build. Record which model and version you deployed, where it runs, what the licence permits, what evaluation you ran on sensitive-topic behaviour, and who signed it off. That record is what turns a defensible decision into an evidenced one, and it takes an afternoon.
