Head to head
QwQ-32B vs Qwen3-32B
QwQ-32B from Alibaba Qwen against Qwen3-32B from Alibaba Qwen — specification, price and every benchmark both makers have published, in one table.
Benchmarks
Qwen3-32B leads
Qwen3-32B wins 3 of the 3 benchmarks both models report, QwQ-32B wins 0, though the gaps are small. The average gap across shared tests is 1.9 points.
Price
Qwen3-32B is cheaper
On a 3:1 input-to-output mix, Qwen3-32B costs $0.150 per million tokens against $0.162 for QwQ-32B — a small difference. Remember that a reasoning model bills its thinking as output, so cost per answer can diverge much further than cost per token.
What actually differs
- These two are closely matched on the specification side — same broad capabilities, same licensing posture. The decision comes down to the benchmark rows and the price.
Scorecard
Which is better at what
Maths, coding, reasoning and the rest — one line each, averaged over the benchmarks both models actually report.
| Category | QwQ-32B | Qwen3-32B | Better at this |
|---|---|---|---|
| ReasoningGPQA Diamond | 65.2% | 66.8% | Qwen3-32B+1.6 |
| MathsAIME 2025 | 79.5% | 81.4% | Qwen3-32B+1.9 |
| CodingLiveCodeBench | 63.4% | 65.7% | Qwen3-32B+2.3 |
| KnowledgeNeither model reports this | — | — | Not comparable |
| MultimodalNeither model reports this | — | — | Not comparable |
| Instruction followingNeither model reports this | — | — | Not comparable |
| Human preferenceNeither model reports this | — | — | Not comparable |
| Categories wonOut of 3 comparable | 0 | 3 | Qwen3-32B |
Each category averages only the benchmarks every model here reports, so no one is credited for a test the other did not run. A category with no shared test is marked Not comparable rather than guessed at.
Side by side
QwQ-32B and Qwen3-32B, row by row
| Attribute | QwQ-32BAlibaba Qwen | Qwen3-32BAlibaba Qwen |
|---|---|---|
| Specification | ||
| MakerWho built it | Alibaba Qwen | Alibaba Qwen |
| Released | 2025-03 | 2025-04 |
| ParametersTotal, and active per token for a mixture of experts | 32B | 32B |
| Architecture | Dense transformer | Dense transformer |
| Context windowHow much can go in at once | 131,072 tokens | 131,072 tokens |
| Max output | 32,768 tokens | 32,768 tokens |
| Input | Text | Text |
| ReasoningSpends extra tokens thinking before it answers | Yes | Yes |
| Tool calling | Yes | Yes |
| Licence | Apache 2.0 | Apache 2.0 |
| Open weightsCan you download and run it yourself | Yes | Yes |
| Price | ||
| Input priceUSD per million tokens in | $0.15 | $0.10 |
| Output priceUSD per million tokens out | $0.20 | $0.30 |
| Blended 3:1A 3-in-to-1-out million-token mix — a fairer single number than input price alone | $0.162 | $0.150Cheapest |
| Price note | Open weights — this is a representative hosting rate, not a first-party price. Running it yourself costs only hardware. | Open weights — this is a representative hosting rate, not a first-party price. Running it yourself costs only hardware. |
| Published benchmarks | ||
| GPQA Diamond198 graduate-level physics, chemistry and biology questions written to be Google-proof. PhD holders in the matching field score about 65%. | ||
| AIME 2025The American Invitational Mathematics Examination — 15 problems, integer answers, no partial credit. A standard test of multi-step maths reasoning. | ||
| LiveCodeBenchCompetitive-programming problems collected after each model's training cutoff, so contamination cannot inflate the score. | ||
| Links | Hugging Face · Full page | Hugging Face · Full page |
| Row verified | 2026-08 | 2026-08 |
Where these numbers come from
Every score on this page is a published figure, taken from the model's own card, system card, technical report or release post, or from a public leaderboard. CorX Labs did not run these evaluations. Most are self-reported by the lab that built the model, which means they were produced under that lab's own choice of prompt, scaffold and number of attempts — so treat them as a starting point for a shortlist, not as a settled ranking.
A score someone other than the model's maker measured is marked Independent and names its measurer. Those are the stronger numbers on this page — an outside harness has no reason to flatter anyone — and there are not many of them.
Where a figure has not been published, the cell reads Not reported rather than an estimate. Nothing here is inferred, interpolated or guessed. Each model records the month its row was last checked. Full method and caveats.
Questions
QwQ-32B or Qwen3-32B?
Is QwQ-32B better than Qwen3-32B?
Qwen3-32B wins 3 of the 3 benchmarks both models report, QwQ-32B wins 0, though the gaps are small. The average gap across shared tests is 1.9 points.
Which is cheaper, QwQ-32B or Qwen3-32B?
On a 3:1 input-to-output mix, Qwen3-32B costs $0.150 per million tokens against $0.162 for QwQ-32B — a small difference. Remember that a reasoning model bills its thinking as output, so cost per answer can diverge much further than cost per token.
What is the difference between QwQ-32B and Qwen3-32B?
These two are closely matched on the specification side — same broad capabilities, same licensing posture. The decision comes down to the benchmark rows and the price.
