AI Tools Index / Open models / Qwen
Alibaba Cloud · reviewed Sep 2026 · assessed from published documentation and our own evaluation tasks

Qwen

Qwen3 generations, including small dense models and large mixture-of-experts releases

The best licence-to-capability ratio in open weights: Apache-licensed for most sizes, strong instruction following, and a small-model range that makes a genuinely private pilot affordable for a mid-sized firm.

Our verdict

Tier B — conditional, and the model we would put in front of a firm that wants a private deployment with the fewest licence complications. Its weaknesses are behavioural rather than structural: it reasons past silence and it is terse. Both are manageable with prompt discipline, both should be tested on your own matters before rollout.

Specifications, as published

PublisherAlibaba Cloud
FamilyQwen3 generations, including small dense models and large mixture-of-experts releases
ParametersSpans small dense models suited to a single workstation through to large mixture-of-experts releases
ContextLong context across the range; small models hold up better than their size suggests on retrieval-grounded tasks
LicenceApache 2.0 for the majority of releases — the most straightforward commercial licence in this index
WeightsDownloadable, with strong quantised community support
ReleaseContinuous release cadence; version pinning is essential
Licence postureApache 2.0 for the majority of releases — the most straightforward commercial licence in this index

Specifications are as published by the publisher at the review date and change frequently. Confirm them in your own evaluation before you procure.

Releases and variants

ReleaseSizeContextServing footprintWhat it is for
Qwen3.5-0.8B0.8B dense, vision-language262,144 tokens natively~0.5GB at 4-bit quantisation; ~1GB at 8-bitThe publisher describes this size as being for prototyping and fine-tuning, and its documented tendency to enter thinking loops means it belongs in experiments and classification pipelines rather than anywhere near client material.
Qwen3.5-2B2B dense, vision-language262,144 tokens natively~1.5GB at 4-bit quantisation; ~2.5GB at 8-bitSmall enough to sit inside a workstation's spare memory for tidy extraction and translation work, but a firm of 10–50 fee-earners should use it as one component in a pipeline rather than as the model answering a fee-earner's question.
Qwen3.5-9B9B dense, vision-language262,144 tokens natively, extensible to 1,010,000 tokens~6GB at 4-bit quantisation; ~10GB at 8-bitThe smallest release in this family we would point at real documents: one 24GB card gives a small firm credible extraction and summarisation, subject to the judgement ceiling that a 9B model implies.
Qwen3.6-35B-A3B35B total / 3B active, 256 experts262,144 tokens natively, extensible to 1,010,000 tokens~18GB at 4-bit quantisation; ~36GB at 8-bitThe throughput play in this family — only 3B parameters active per token, so a single 24GB card keeps a busy team served on extraction and triage where a dense 27B would queue.
Qwen3.8-27B27B dense, vision-language, thinking mode on by default262,144 tokens natively, extensible to 1,000,000 tokens~15GB at 4-bit quantisation; ~28GB at 8-bitOur recommended default: a firm of 10–50 fee-earners can serve it at 8-bit on one 48GB card, and it combines current-generation capability with the most straightforward Apache 2.0 terms in this index.
Qwen3.5-122B-A10B122B total / 10B active, 256 experts262,144 tokens natively, extensible to 1,010,000 tokens~63GB at 4-bit quantisation; ~125GB at 8-bit; ~250GB for the published BF16 weightsA realistic step up for a firm that has outgrown the 27B release — an 80GB card runs it at 4-bit and the 10B active parameters keep latency civil — but the quality gain has to be proved on your own matters before you buy the card.
Qwen3.8-Flash-Next125B total / 6B active, plus 51B n-gram embedding and 4B multi-token-prediction module262,144 tokens natively, extensible to 1,000,000 tokens~65GB at 4-bit quantisation for the MoE weights, plus ~50GB of n-gram embedding tables that the publisher notes can be offloaded to host memoryThe low-latency, long-context release built for high-volume serving on an 80GB card or more, and the one to read the licence on: Qwen Community Licence 1.0 permits a firm's internal use but not offering the model's capabilities to third parties.
Qwen3.8-2.4T-A95B2.4T total / 95B active, 512 experts (10 routed plus 1 shared)262,144 tokens natively, extensible to 1,010,000 tokens~1.2TB at 4-bit quantisation; ~4.9TB for the published BF16 weightsCapable and impractical for a 20-partner firm: 2.4T parameters is a data-centre deployment under a bespoke Qwen3.8-Max licence, and a firm of this size should rent capacity or use a hosted endpoint rather than plan to serve it.

Sizes, context windows and licences are as published by the publisher at the review date. The variant you pick matters more than the family name: a small dense release that fits one workstation and a large mixture-of-experts release that needs a multi-GPU server are not the same product, whatever the marketing says.

How it behaves on legal work

For legal work Qwen's most useful trait is instruction obedience. Give it a schema, a tone, a length limit and a refusal rule, and it will follow all four — which sounds unremarkable until you have watched other models quietly ignore the fourth. Its extraction behaviour is the strongest of the models we have assessed: asked to pull dates, parties, obligations and termination rights out of a contract set into a fixed structure, it produces clean output that a paralegal can validate rather than repair. Drafting is competent and slightly terse; letters come out shorter and less padded than Llama's, which many fee-earners prefer and some partners find abrupt. On research-style questions it is fast and structured but has a visible weakness: it will happily reason from its own general knowledge when retrieval returns nothing useful, and its confident phrasing in those moments is the risk. Handling of Chinese-language material is an obvious strength, which matters for firms with cross-border matters and is a genuine differentiator against most of this index. The multi-step reasoning mode is worth using for triage — allocation of documents to issues, or deciding what to read next — rather than for final legal conclusions.

evidence and abstention

Abstention is good when a refusal rule is explicit and the retrieved context is empty. It is average when the context contains adjacent-but-not-answering material: it will bridge the gap rather than flag it. Quote-with-reference instructions work well; ask for paragraph-level citation and it complies.

What we would use it for

  • Contract extraction into a fixed schema at volume
  • Disclosure triage and document allocation to issues
  • Multilingual matter support, including Chinese-language documents
  • Drafting from precedent where brevity is wanted
  • A genuinely affordable first private pilot on a single GPU

What to watch

  • Reasoning past the retrieved evidence when the corpus is silent
  • Version churn — pin the exact release in your evaluation record
  • Provenance questions from clients: be ready to explain the publisher's jurisdiction and terms
  • Terse drafting that can read as blunt in client-facing correspondence

What it costs to run

We would serve Qwen3.8-27B at 8-bit on a single 48GB card: that preserves the long context the family advertises, leaves room for the KV cache of several concurrent requests, and keeps the deployment on plain Apache 2.0 terms. For a firm of 10–50 fee-earners that is comfortable interactive latency, because schema-bound extraction is largely input-bound and parallelises well, while long drafting runs come back nearer reading speed. Where throughput rather than quality is the constraint, Qwen3.6-35B-A3B at 4-bit runs on a 24GB card with 3B parameters active per token; the 122B and 2.4T releases are separate purchases with separate hardware consequences.

BasisGPU hoursHourly (USD)Monthly (USD)When this is the right pattern
Always-on server (24/7)730 h$0.85 – $2.85$625 – $2,080Firm-wide access, no cold starts, predictable latency
Business hours (10 h × 21 days)210 h$0.85 – $2.85$180 – $600The realistic pattern for a firm of 10–50 fee-earners: power it up, use it, stop paying
Bursty / autoscaled endpoints60 h$1.80 – $3.60$110 – $215Occasional analysis and pilots; you pay only for the seconds the model is working
Storage — weights, index and evaluation sets (~200 GB)$30Billed whether the model is running or not — the quiet line on the invoice

Indicative GPU class: 48GB class — L40S / RTX A6000. Every figure above includes a 50% buffer on the underlying cloud rates — for encrypted storage, egress, idle capacity between requests, cold starts, operational overhead, and the plain fact that these are estimates rather than quotes. Rates move weekly and vary by region, tier and commitment.

the comparison that decides it

Qwen is the family where buying hardware is most defensible, because the 27B release we would serve fits a single 48GB-class workstation — indicatively $8,000–15,000 for the class — and the Apache 2.0 terms let you leave it in place for years without a licence conversation. Renting the same class remains the right first step if your usage amounts to a few hours a week, since idle capital is the largest avoidable cost in private AI. Those bands are indicative rather than quotes and take no account of the retrieval layer and evaluation time that will dominate your real spend; the 122B release above needs an 80GB-class server, which is a materially larger commitment.

Two rules of thumb that hold across the models we have deployed: renting beats buying until a firm is using the model more than about half of every working day, and stopping the instance matters more than the hourly rate — an idle server, and an idle storage volume attached to it, are where private AI budgets quietly go.

Sampling and prompt settings

Extraction: temperature 0, top_p 0.8, fixed schema, no creative latitude. Triage: 0.2–0.3. Drafting: 0.4. Research summaries: 0.2 with an explicit 'cite the passage' instruction. The small variants can be pushed to 0 for classification without collapsing into repetition.

Pin the exact model release in your evaluation record. Behaviour moves between point releases, and an evaluation that does not name a version cannot be reproduced.

Hardware and quantisation

Deployment profileWhat it fitsWhat to know
Single workstation (small dense, quantised)Extraction and summarisation for one teamCheapest credible private pilot in this index
One GPU server (mid-size)Firm-wide extraction and drafting supportGood throughput per pound for schema-bound work
MoE release, larger deploymentResearch-style tasks and heavier drafting volumeEvaluate against the mid-size model before paying for the step up

Who it suits

Good fit

Firms wanting Apache-licensed capability, cross-border and multilingual work, and high-volume extraction where an output schema matters more than elegant prose.

Poor fit

Firms that need warm client-facing drafting out of the box, or that cannot commit to pinning and re-evaluating model versions.

Review history

DateChange
Sep 2026First entry.

Sources

Published under our rubric. Specifications are as published by the model publisher at the review date; licences and capabilities change without notice, so verify before you procure. Scores are editorial opinion formed from published documentation and our own evaluation tasks — not a benchmark result and not a vendor statement. No publisher pays for placement, sees a score before publication, or can have an entry withdrawn. Nothing here is legal advice; test any model on your own matters before you put client data through it.