Get started

Qwen3 32B

Sold by 5 sellers, timed at 29.1 tok/s, scored 15.3 on the Coding Index, and with results on 2 openly licensed boards.

retrieved sha256 · 0bb53a77713f

Models/Qwen/Qwen3 32BUSD per 1M tokens · as OpenRouter routes it
qwen/qwen3-32bQwen · base listing · Qwen3 tokenizer
OpenRouter list price · the routed rate, as the catalog carried itretrieved 17 Nov 16:29 UTC from openrouter.ai ↗
Input$0.08per 1M tokens
Output$0.28per 1M tokens
Cached inputnot listedper 1M tokens
Context131k131,072 tokens
What Vitna read · 2 providers, each from its own pricing page, kept as dated recordsin / out per 1M tokens
AWSopen weights$0.15 / $0.60confirmed 22 Nov 2026
DeepInfraopen weights$0.08 / $0.28confirmed 22 Nov 2026

The same weights served by 2 providers, each at its own list price. OpenRouter's rate is its routed rate across whichever of them it uses, and it need not match any one of them.

What the list says · the fields OpenRouter publishes for this id11 fields
AuthorQwen
Listed
Canonical idqwen/qwen3-32b-04-28
Context131,072 tokens
Max output16,384 tokens
Modalitiestext → text
TokenizerQwen3
Reasoningoptional
Moderationnot moderated by OpenRouter
Knowledge cutoff2025-03-31
What the weights say · read from the model's own repository on Hugging Faceretrieved from huggingface.co ↗
Parameters32.8B · 32,762,123,264 declared
Model classQwen3ForCausalLM
Downloadnot gated
Published

What sellers charge

5 sellers · 2.19x from cheapest to dearest

One tick per seller, on a log scale of input price per million from $0.064 at DeepInfra to $0.14 at Tenstorrent. 1 seller undercuts the $0.08 listed; 3 are dearer. A faint tick declares no number format. Every seller is named on the endpoint ledger below.

Measured speed

19,876 requests · 5 endpoints · 30-minute window
Throughput29.1 tok/sMedian output rate, weighted across the 5 endpoints serving it by the requests each answered.the speed board
First token405msMedian wait before the first token, over the same window and weighted the same way.how Vitna reads a figure
Endpoint spread29 to 34The slowest and the fastest route serving this model, in tokens per second. Speed is a property of the route, not of the weights.the board, fastest first

OpenRouter's own timing of its routed traffic, weighted by requests, over one 30-minute window read . Every route is on the endpoint ledger below.

Every endpoint

5 priced · 5 timed · cheapest first
  1. DeepInfradeepinfra/fp8/flex$0.064$0.224not listedfp837knot matchednonenonenot matched66.66%
  2. DeepInfradeepinfra/fp8 · listed$0.08$0.28not listedfp816knot matchednonenonenot matched99.77%
  3. DeepInfradeepinfra/fp8/priority$0.12$0.42not listedfp837knot matchednonenonenot matched100%
  4. SiliconFlowsiliconflow/fp8$0.14$0.57not listedfp8118k30.5 tok/s261.52s304 req100%
  5. Tenstorrenttenstorrent/fp8$0.14$0.40$0.09fp816k31.5 tok/s8382ms464 req92.53%
  6. DeepInfranot matchednot matchednot matchednot matchedno offer34 tok/s22.81.35s1 reqno offer
  7. DeepInfranot matchednot matchednot matchednot matchedno offer29 tok/s21446ms4,223 reqno offer
  8. DeepInfranot matchednot matchednot matchednot matchedno offer29 tok/s20372ms14,884 reqno offer

Each track runs from a route's p90, the rate nine in ten of its requests met or beat, up to its median. Weighted by requests, the figure above reports where the traffic went rather than what one route does: 2 of the 5 timed routes are slower than it, and the middle route runs at 30.5 tok/s. The widest median-to-p90 gap is Tenstorrent's, 3.94x.

DeepInfra publishes 3 offers and had 3 routes timed. The two documents share no route id, so those offers and routes are printed apart rather than paired.

Uptime is OpenRouter's own figure for each seller over the trailing day, read once a day, and the column prints the latest reading, taken . 4 of the 5 sellers with a reading fell under 99% on at least one of the 23 days held, 23 Oct 2026 to 21 Nov 2026: Tenstorrent at 63.79% on 3 Nov 2026, DeepInfra (deepinfra/fp8/flex) at 66.66% on 21 Nov 2026, SiliconFlow at 81.38% on 19 Nov 2026, and DeepInfra (deepinfra/fp8) at 91.66% on 23 Oct 2026. Source: OpenRouter's public endpoints API, one reading a day for each seller: 1,211 sellers carried one on the latest of the 23 days held.

Source: OpenRouter's public endpoints API. 1257 sellers across 359 models when the mirror ran, 186 of them served by more than one. List prices, not what anybody paid; a discount is the seller's own. Read . Speed is OpenRouter's own timing over one 30-minute window, read .

What each seller accepts

20 parameters · 5 sellers
  1. frequency_penaltyevery seller
  2. include_reasoningevery seller
  3. max_tokensevery seller
  4. reasoningevery seller
  5. temperatureevery seller
  6. top_kevery seller
  7. top_pevery seller
  8. logit_bias4 of 5not SiliconFlow
  9. min_p4 of 5not SiliconFlow
  10. presence_penalty4 of 5not SiliconFlow
  11. repetition_penalty4 of 5not SiliconFlow
  12. response_format4 of 5not Tenstorrent
  13. seed4 of 5not SiliconFlow
  14. stop4 of 5not SiliconFlow
  15. structured_outputs4 of 5not Tenstorrent
  16. tool_choice3 of 5not DeepInfra (deepinfra/fp8/flex) and DeepInfra (deepinfra/fp8/priority)
  17. tools3 of 5not DeepInfra (deepinfra/fp8/flex) and DeepInfra (deepinfra/fp8/priority)
  18. logprobs1 of 5only Tenstorrent
  19. top_a1 of 5only Tenstorrent
  20. top_logprobs1 of 5only Tenstorrent

7 of these 20 are accepted by every one of the 5 sellers that publish a list; the other 13 are declared by some sellers and not others, so what a request sending one of them gets depends on which seller serves it. Each list is the seller's own declaration in OpenRouter's endpoints API, read , and nothing here tested one.

Price position

342 priced models

Blended $0.13 per million tokens, 3 in to 1 out: cheaper than 297 of the 342 priced models in the catalog. Each bar is one stretch of the catalog on a logarithmic price scale; the mark is this listing.

Measured quality

145 models with an index · 132 on a design arena board

Each mark is one board this model stands on, placed at the share of that board it scores above, and led to its number in the list under the rule. The columns and the ledger below carry the figures.

Intelligence Indexnot scoredArtificial Analysis has not scored this model on its Intelligence Index.the intelligence leaderboard
Coding Index15.3Position 103 of 118, above 13% of the models scored.the coding leaderboard
Agentic Index0.9Position 85 of 94, above 9% of the models scored.the agentic leaderboard

Not ranked in any Design Arena category.

Epoch Capabilities Index · a fit by a third party over every benchmark this model has results onthe whole index, with intervals
Index136.8134.4 to 139.1
Position136 of 258above 47% of the models placed
Results behind it6benchmark results
As Epoch names itQwen3-32Breleased 29 Apr 2025

Position 136 of the 258 models Epoch has placed, which is not the population of either board above, on a scale anchored at Claude 3.5 Sonnet 130 and GPT-5 150. Epoch states that absolute values are meaningless by themselves and that comparisons between models are what the index is for, that values are not linearly related to benchmark accuracy, and that a highly specialised model may score low while being capable in its own domain.

Benchmark results

2 of 6 boards · 5 results
Epoch AI Benchmarking Hub · publisher-run, CC-BY-4.02 of 22 tasks
TaskAll runsEffortScore
GPQA diamond342none0.54
unstated0.66
OTIS Mock AIME 2024-2025319none0.23
unstated0.67

Mean accuracy, 0 to 1, on the task the row names, as Epoch AI runs it and averaged over repeated runs. All eighteen of their tasks, including the held-out FrontierMath tiers, named as they name them. The model id carries the effort setting the run used. Epoch AI Benchmarking Hub names these runs Qwen3-32B_none and Qwen3-32B, and the effort beside each score says which run it is, unstated where the publisher wrote none. All runs is how many runs the publisher scored on the task, every model's. Read from epoch.ai ↗, which dates its own data . Epoch AI Benchmarking Hub data, by Epoch AI, under CC BY 4.0.

Aider polyglot · submitted, Apache-2.01 of 1 task · 69 runs scored on the task · effort unstated

Percent of 225 Exercism exercises solved with a second attempt allowed, higher is better. The id carries the edit format, because the publisher says a model alone does not identify a run. Aider polyglot names this run Qwen3 32B (diff), with no effort setting written on it. Read from raw.githubusercontent.com ↗, which dates its own data . Aider polyglot benchmark, Apache-2.0, by the Aider project. Results are submitted by contributors rather than produced by the publisher.

2 of the 6 openly licensed boards Vitna mirrors carry a result for Qwen3 32B, each printed as its publisher published it and none added to another. MTEB, LiveBench, HELM and Vectara hallucination leaderboard have not published one under a name that matches its listing, and names are matched to the catalog exactly and never guessed. A task that is a link opens its column on /capability, with every run scored on it; an effort that is a link opens its run's row there, where the run is ranked on Vitna's scale. Every board, with its licence and what its publisher measures, is on /benchmarks.

Sources

Source: OpenRouter (openrouter.ai/models), as of 17 Nov 16:29 UTC. sha256 0bb53a77713f over 853,270 bytes.
Catalog: OpenRouter's public model list.
Intelligence, coding and agentic indexes: Artificial Analysis, as published by OpenRouter.
Elo, win rate and rank: Design Arena, as published by OpenRouter.
Source: each model's own Hugging Face repository (huggingface.co), as of 20 Nov 2026. 162 repositories read, 4 did not answer.
Licence, parameter count, model class and gating: each model's own Hugging Face repository, the one OpenRouter names for the listing.
A licence is the tag the uploader put on the repository, read from nothing else and verified by nobody.
A parameter count is the total the repository's safetensors files declare; for a mixture of experts that is every expert, not the share one token runs through.
Source: the Epoch AI Capabilities Index (epoch.ai/eci), as of 20 Nov 2026. 258 models placed over 2917 benchmark results.
Epoch AI Benchmarking Hub data, by Epoch AI, under CC BY 4.0.
Vitna ran none of these evaluations and adjusted none of these figures.