NOUMENOpen app
NOUMEN

Qwen · Open weights

Qwen3 VL 235B A22B Thinking

Added Aug 26, 2025Open weights

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Use in the app

Qwen3 Vision‑Language model built on a 235B MoE backbone (≈22B active per token). Strong at OCR, charts/tables, multi‑image reasoning, and complex document understanding. The Thinking variant enables long‑form, chain‑of‑thought style reasoning.

  • Reasoning
  • Vision
Input
Text, Images
Output
Text
Discovery tier
Open weights

Pricing

USD per 1M tokens
Input
$0.50
Output
$6.00

Reference prices as published by the model provider. API example

Specifications

Context window
—
Max output
32.8K

Side by side

Qwen3 VL 235B A22B Thinking compared

Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Qwen3 VL 235B A22B ThinkingQwenThis modelNvidia Nemotron 3 Ultra 550BnvidiaRemove
ProviderQwennvidia
Discovery tierOpen weightsOpen weights
AddedAug 26, 2025Jun 4, 2026
Limits
Context window—1M
Max output32.77K65.54KBest
Average output—56 tokens
Pricing, USD per 1M tokens
Input$0.50$0.50
Output$6.00$2.50Best
Performance
Throughput—92.8 tok/s
Time to first token—1.1 s
Request success (24h)—100.0%
Benchmark indexes
Intelligence Index—22.9
Coding Index—49.3
Capabilities
ReasoningYesYes
ToolsNoYes
VisionYesNo
Parallel toolsNoYes
Reasoning effort——

Public evaluations

Benchmarks

Independent evaluation results, with how each score ranks among the models in the Noumen catalog.

  • Qwen3 VL 235B A22B Thinking
  • Nvidia Nemotron 3 Ultra 550B

Intelligence Index

  • Qwen3 VL 235B A22B Thinking—
  • Nvidia Nemotron 3 Ultra 550B22.9
3.8Catalog models by score55.1

Indexes

  • Coding Index

    Composite of the coding evaluations

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 49.3

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 3.0%

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 0.3%

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 876 Elo

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 1,013 Elo

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 5.0%

  • AA-LCRv1.1

    Reasoning across long documents

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 79.3%

  • MLCR-AA

    Long-context reasoning over medical records

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 5.6%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 28.4%

  • GPQA Diamond

    Graduate-level science questions

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 86.7%

  • IFBench

    Following precise output instructions

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 81.4%

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 0.5%

  • SciCode

    Scientific computing in Python

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 40.3%

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 36.4%

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 83.3%

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Qwen3 VL 235B A22B Thinking: —
    • Nvidia Nemotron 3 Ultra 550B: 79.3%

Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis

Developers

Use it through the API

Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.

Read the quickstart
curl https://www.noumen.xyz/api/v1/chat/completions \
  -H "Authorization: Bearer $NOUMEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen/qwen3-vl-235b-a22b-thinking",
    "messages": [{ "role": "user", "content": "Explain this stack trace." }]
  }'
import os
from openai import OpenAI

client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
    model="qwen/qwen3-vl-235b-a22b-thinking",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

Qwen

Other Qwen models, starting with the same family.