Skip to content
ModelCap

Qwen3 235B A22B Instruct 2507

Current board #34Arena #103Open weights

Qwen·qwen/qwen3-235b-a22b-2507

Qwen3 235B A22B Instruct 2507 is a language model published by Qwen. 10 OpenRouter endpoints are published in the latest snapshot. It accepts text input and returns text. The published context limit is 262K tokens. The catalogue declares support for tool use, structured outputs.

ModelCap Index · Current board
70.5#34
Measured · 86% support · score interval 70.5–76.9
Market Gravity
41.0
Secondary market signal
Input
$0.09
per 1M tokens
Output
$0.55
per 1M tokens
Context
262K
tokens
Providers
10
endpoints
Gateway spend
—
latest · Vercel share

Providers

Uptime measured over the last 30 minutes
ProviderInputOutputContextQuantUptime
DeepInfrafp8Cheapest$0.09$0.55262Kfp898.49%
Novitafp8$0.09$0.58131Kfp898.49%
Alibabafp8$0.15$0.598131Kfp8100.0%
Nebiusfp8$0.20$0.60262Kfp890.22%
Venicefp8$0.15$0.75128Kfp896.75%
Parasailfp8$0.14$0.80131Kfp8—
Crusoebf16$0.22$0.80262Kbf1699.73%
StreamLake$0.21$0.84128K—99.61%
AtlasCloudfp8$0.20$0.88131Kfp899.45%
Googleus-south1$0.22$0.88262K—100.0%

Hugging Face Inference Providers

Exact repository identity

Provider economics and telemetry published by the Hugging Face Router. These are operational signals, not benchmark results.

ProviderInputOutputContextTTFTThroughputFeatures
deepinfra
live
$0.09$0.55262K545 ms41.2 tok/stools · structured
novita
live
$0.09$0.58131K702 ms47.8 tok/stools · structured
nscale
live
$0.20$0.6033K615 ms13.5 tok/stools · structured
scaleway
live
$0.855$2.56—441 ms54.5 tok/stools · structured

Reported evaluation leads (1)

Excluded from score

Structured rows reported in this exact model repository at the pinned revision below. They are useful research leads, but remain quarantined until dataset revision, harness, configuration and identity pass ModelCap’s independent admission review.

Dataset / taskMetricValueRevisionProvenance
TIGER-Lab/MMLU-Pro
mmlu_pro
—83
Unpinned
not published
Reported
Quarantined

ModelCap Index · Current board

Methodology
70.5
Measured · 3 public benchmark observations across 3 boards
Current board
#34
Public rank basis
Measured (measured)
Index support
85.6% · strong
Index interval
70.5–76.9
Rank posterior
Not available
Top 5 / Top 10 probability
Not available / Not available
Posterior as of
Snapshot timestamp unavailable
Identity binding
Exact catalogue product · qwen/qwen/qwen3-235b-a22b-2507 · aggregates disclosed benchmark configurations · endpoint configuration not claimed · artifact metadata Qwen/Qwen3-235B-A22B-Instruct-2507@ac9c66cc9b46af7306746a9250f23d47083d689e (not the evaluation revision)
Direct benchmark observations
3
Breadth-adjusted BES point estimate
73.7
Bayesian Evidence Score · measured evidence input only

BES normalizes admitted public benchmark evidence for measured Index rows. Its legacy score and observables prior do not define the public language rank.

Observed capability
73.7
Evidence status
confirmed · 94% mass
General preference
73.5
Coding
74.8
Agents & tools
73.9
Reasoning
—
Evidence breadth
97%
Evidence coverage
86%
Market and catalogue signals · secondary, never the language Index rank
Usage (OpenRouter popularity)
26.7
Liquidity (providers × uptime)
86.8
Open reach (HF downloads)
22.8
Surface (context / tools / modalities)
50.3
Freshness
23.3
  • · OpenRouter popularity #65
  • · 10 live providers
  • · HF 30d downloads 56086
  • · Catalogue: context 262144, tools
  • · First seen 2025-07-21

These adoption and deployment observations remain context only. They do not change this model's ModelCap Index score or rank.

BES evidence selected · measured input only

The bounded 0–100 capability score blends each source’s competitive placement with its published achievement, then combines capability families. Evidence breadth adds a modest uncertainty adjustment; price and popularity are not part of benchmark-led rank.

Deployment readiness

Metadata index
90.6
strong · 94% metadata coverage
Not capability
Artifact reproducibility
85.0
Access & legal clarity
100.0
Deployability
99.6
Evaluation provenance
76.1

Missing: artifactReproducibility.base-lineage, evaluationProvenance.evaluation-dataset-revisions.

A metadata completeness and deployability index—not a safety certification, quality grade, or production-readiness claim.

Market activity

Venue coverage
Market Gravity
41.0
OpenRouter weekly popularity
#65 of 298

Market Gravity uses OpenRouter’s full-catalogue popularity order. Sparse Vercel top-ten observations appear only when published and do not affect the score.

Arena preference

Status
Ranked
Source
Arena
Official rank
#103 of 385
Rating
1422.9
95% confidence interval
1420–1425
Votes
97K
Source snapshot
6 Aug 2026, 13:00 UTC
Category
overall
Identity match confidence
100%

Market Gravity breakdown

Usage55%
26.7

OpenRouter weekly popularity across the full model catalogue.

Liquidity25%
86.8

Independent providers versus the model's open or closed cohort, weighted by uptime.

Open reach15%
22.8

Hugging Face 30-day downloads for open models; neutral for closed models.

Freshness5%
23.3

Time since first public availability, on a six-month half-life.

Specification

Context window
262K tokens
Max output
16K tokens
Inputs
Text
Outputs
Text
Tokenizer
Qwen3
Cached input tokens
Not offered
First seen on OpenRouter
21 Jul 2025
Knowledge cutoff
2025-06-30

Capabilities

  • Not supported: Reasoning
  • Supported: Tool use
  • Supported: Structured output
  • Supported: Response format
  • Not supported: Moderated

Weights & access

Open weights. Downloadable weights are published under an OSI or free-culture license in the current repository metadata. This describes weight availability, not whether the full training stack qualifies as Open Source AI.

Access
Open weights
Licence
apache-2.0
Pinned revision
ac9c66cc9b46
Parameters
235.1B
Architecture
Qwen3MoeForCausalLM
Model type
qwen3_moe
Hugging Face downloads (30d)
56K
Downloads (all time)
1.5M
Likes
793
Repository
Qwen/Qwen3-235B-A22B-Instruct-2507
Repository updated
17 Sept 2025, 06:52 UTC