Skip to content
aiqbench

Alibaba · model passport

Qwen: Qwen3 VL 235B A22B Instruct

qwen3-vl-235b-a22b-instruct
  • Open weights
  • Apache 2.0 licence
  • 235.7B params
  • 262,144 context
  • Released 2025-09-22(HF repo created)
9/27axes covered
Overall · LMArena
1,413#133 of 396
Best percentile
79.8MMLU-Pro (Artificial Analysis)
Cohort standing
57.6median 52.6

What the data says

Benchmarked on 9 of 27 tracked axes. Every figure below is the model's own published result on that board.

Uneven: 55.4 percentile points between its best and worst axis — 79.8 pct on MMLU-Pro (Artificial Analysis) versus 24.4 pct on Factuality Arena.

  • MMLU-Pro (Artificial Analysis)79.8 pct82.3%Verify →
  • Instruction following70.2 pct1,409Verify →
  • Coding69.3 pct1,463Verify →
  • Hard prompts69.2 pct1,438Verify →
  • Overall66.7 pct1,413Verify →
  • LiveCodeBench (Artificial Analysis)63 pct59.4%Verify →
  • Vision Arena52 pct1,215Verify →
  • IFBench (Artificial Analysis)46.3 pct42.7%Verify →
  • Factuality Arena24.4 pct1,414Verify →
  • WebDev ArenaNot benchmarked
  • Search ArenaNot benchmarked
  • Document ArenaNot benchmarked
  • Agent ArenaNot benchmarked
  • GPQA DiamondNot benchmarked
  • AIME (math)Not benchmarked
  • FrontierMathNot benchmarked
  • SimpleQA VerifiedNot benchmarked
  • ARC-AGI-2Not benchmarked
  • Humanity's Last ExamNot benchmarked
  • SimpleBenchNot benchmarked
  • LiveBenchNot benchmarked
  • BFCL v4 (function calling)Not benchmarked
  • SWE-bench Verified (live board)Not benchmarked
  • Terminal-Bench 4.0Not benchmarked
  • HHEM Hallucination (Vectara)Not benchmarked
  • SWE-bench VerifiedNot benchmarked
  • Terminal-BenchNot benchmarked

Percentiles are the model's standing within each board's own population. Axes marked "not benchmarked" have no published result for this model — that is an absence of data, not a low score.

How to run it

Run it yourself

RepositoryQwen/Qwen3-VL-235B-A22B-Instruct
Parameters235.7B
ArchitectureQwen3VLMoeForConditionalGeneration
LicenceApache 2.0

Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "Qwen/Qwen3-VL-235B-A22B-Instruct"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)

Ollama

We don't sync Ollama tags, so we won't print a command we can't verify — a wrong one would pull the wrong model. Search the Ollama library →

Unsloth fine-tuning recipes are model-specific and we don't derive them here. Unsloth docs →

Use it hosted

API model IDqwen/qwen3-vl-235b-a22b-instruct
Context length262,144 tokens
Max output32,768 tokens
Acceptstext, image
Input price$0.21 per 1M tokens
Cached input price$0.10 per 1M tokens
Output price$1.90 per 1M tokens

API parameters this endpoint supports

Frequency penaltyLogit biasLogprobsMax tokensMin-pPresence penaltyRepetition penaltyResponse formatSeedStop sequencesStructured outputsTemperatureTool choiceTool callingTop-kTop logprobsTop-p

Example request

curl https://openrouter.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"qwen/qwen3-vl-235b-a22b-instruct","messages":[{"role":"user","content":"Hello"}]}'

View on OpenRouter →

Identifiers, limits and prices are synced from the upstream source shown above. Any field we haven't synced is left out entirely rather than guessed.

  • qwen3-vl-235b-a22b-instruct stands at 57.6 (raw mean 60.1) — measured on 9 of 27 axes.

    standingtemplateVerify →

  • qwen3-vl-235b-a22b-instruct stands at 57.6 versus a cohort median of 52.6 across 143 models released within 180 days — above the peer median.

    cohorttemplateVerify →

Every claim is a computed fact from the current board data, phrased by a low-temperature model call and rejected by a validation gate if any number is not in the source row. How the leaderboards work →