Inference
EN
Alibaba

Qwen3 235B 2507 (Reasoning)

Qwen3 235B 2507 (Reasoning) is a reasoning model from Alibaba in the Qwen3 235B 2507 family. 8 benchmarks count toward its index, in 5 categories.

  • Open weights
  • Reasoning
  • Input: text
  • Released 2025-07
RankedEnough results for an index.
IndexAll categories, out of 100.47.4±10.2Median 51.0
CoverageIndex weight that has results.55%
SpeedOutput tokens per second.108/sMedian 67/s
First tokenSeconds to the first output token.1.1 sMedian 1.3 s
Input / 1MUS dollars per 1M input tokens.$0.45Median $0.60
Output / 1MUS dollars per 1M output tokens.$3.50Median $2.55
ContextMaximum tokens in one request.128KMedian 262K
EloLMArena rating and rank.1416±7 #133Median 1422

Details

From OpenRouter
API IDThe model ID on OpenRouter.
qwen/qwen3-235b-a22b-thinking-2507
Max outputMaximum output tokens per request.
16K
Knowledge cutoffLast date of training data.
30 Jun 2025
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
No
Cache read / 1MUS dollars per 1M cached tokens.
—
WeightsLink to the published weights.
Hugging Face (opens in a new tab) ↗
RetiresDate OpenRouter removes the model.
—

Where to run

APIs that serve this model.
2 providers
ProviderTypeMaker, cloud, host or router.InputOutputContext
IO.NETDeprecatedHost$0.11$0.60262K
OpenRouterRouter$0.45$3.50128K

US dollars per 1M tokens, list price. From models.dev (opens in a new tab) (MIT).

Run locally

235.09B parameters
16-bit
494.7 GB
8-bit
263.3 GB
4-bit
150.6 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy needs more than 128 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.None counted——
CodingCode writing and repair.1 counted45.3—
ReasoningLogic problems and puzzles.None counted——
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.4 counted48.9142 / 268
MultilingualTasks in many languages.1 counted50.4—
InstructionTasks with strict prompt rules.1 counted40.0—
MathMath problems.1 counted49.1—

Results

8 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
American Invitational Mathematics Examination 2025Math92.3%14 / 53
Massive Multitask Language Understanding ProfessionalKnowledge84.4%24 / 75
Graduate-Level Google-Proof Q&AKnowledge81.1%63 / 141
Instruction-Following EvalInstruction87.8%32 / 70
LiveCodeBench v6Coding74.1%36 / 59
MMLU-ProXMultilingual81.0%11 / 17
SuperGPQA: Scaling LLM Evaluation Across 285 Graduate DisciplinesKnowledge64.9%16 / 22
Humanity's Last ExamKnowledge18.2%67 / 77

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reusemodels.devMIT

History

5 releases
Index48.3 → 47.4
23 Oct 202612 Nov 2026
Output / 1M$3.50 → $3.50
23 Oct 202612 Nov 2026

Questions

How much does Qwen3 235B 2507 (Reasoning) cost?
$0.45 per 1M input tokens, $3.50 per 1M output tokens.
How good is Qwen3 235B 2507 (Reasoning)?
Its index is 47.4 out of 100. That is place 165 of 283 ranked models.
What is Qwen3 235B 2507 (Reasoning) best at?
Knowledge: 48.9 out of 100, place 142 of 268.
How fast is Qwen3 235B 2507 (Reasoning)?
108 output tokens per second, 1.1 s to the first token.
What is the context window of Qwen3 235B 2507 (Reasoning)?
128K tokens.
Is Qwen3 235B 2507 (Reasoning) open source?
Yes. Its weights are published.
When was Qwen3 235B 2507 (Reasoning) released?
July 2025.

Badge

Qwen3 235B 2507 (Reasoning) on Inference360: place 165 of 283, index 47.4
HTML
<a href="https://inference360.ai/models/qwen3-235b-2507-reasoning"><img src="https://inference360.ai/badges/qwen3-235b-2507-reasoning.svg" alt="Qwen3 235B 2507 (Reasoning) on Inference360: place 165 of 283, index 47.4" height="20"></a>
Markdown
[![Qwen3 235B 2507 (Reasoning) on Inference360: place 165 of 283, index 47.4](https://inference360.ai/badges/qwen3-235b-2507-reasoning.svg)](https://inference360.ai/models/qwen3-235b-2507-reasoning)

Compare with

Gemma 4 31B47.9Qwen3 235B 250742.3dots3-note Preview63.4

Lower price

dots3-note Preview63.4 · FreeApodex 1.1 Mini55.7 · FreeNemotron 3 Ultra50.8 · Free

More from Alibaba

All models
Qwen3-Coder 480B-A35BUnrankedQwen3 235B 250742.3Qwen3 30B A3B Instruct 2507UnrankedQwen3 Coder 30B A3B InstructUnrankedQwen3 30B A3B Thinking 2507Unranked