NOUMENOpen app
NOUMEN

Qwen · Balanced

Qwen3 Max

Added Sep 5, 2025

Compare models

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

Loading models…

Use in the app

Qwen3 Max improves accuracy in coding and science, instruction following, and tool calling.

Input
Text
Output
Text
Discovery tier
Balanced

Pricing

USD per 1M tokens
Input
$1.20
Output
$6.00

Reference prices as published by the model provider. API example

Specifications

Context window
256K
Max output
32.8K

Side by side

Qwen3 Max 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 MaxQwenThis modelGemini 3.1 Flash LiteGoogleRemove
ProviderQwenGoogle
Discovery tierBalancedEfficient
AddedSep 5, 2025Mar 3, 2026
Limits
Context window256K1.05MBest
Max output32.77K65.54KBest
Average output—669 tokens
Pricing, USD per 1M tokens
Input$1.20$0.25Best
Output$6.00$1.50Best
Cache read—$0.025
Cache write—$0.0833
Performance
Throughput—461.8 tok/s
Time to first token—2.4 s
Request success (24h)—100.0%
Benchmark indexes
Intelligence Index15.615.6
Coding Index—34.7
Math Index80.7—
Capabilities
ReasoningNoYes
ToolsNoYes
VisionNoYes
Structured outputNoYes
PDF inputNoYes
Audio inputNoYes
Video inputNoYes
Reasoning effort—minimal, low, medium, high

Public evaluations

Benchmarks

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

  • Qwen3 Max
  • Gemini 3.1 Flash Lite

Intelligence Index

  • Qwen3 Max15.6
  • Gemini 3.1 Flash Lite15.6

Qwen3 Max: Better than 39% of 326 scored models

3.8Catalog models by score55.1

Indexes

  • Coding Index

    Composite of the coding evaluations

    • Qwen3 Max: —
    • Gemini 3.1 Flash Lite: 34.7

  • Math Index

    Composite of the math evaluations

    • Qwen3 Max: 80.7
    • Gemini 3.1 Flash Lite: —

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Qwen3 Max: —
    • Gemini 3.1 Flash Lite: 6.8%

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Qwen3 Max: —
    • Gemini 3.1 Flash Lite: 0.5%

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Qwen3 Max: —
    • Gemini 3.1 Flash Lite: 207 Elo

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Qwen3 Max: —
    • Gemini 3.1 Flash Lite: 430 Elo

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Qwen3 Max: —
    • Gemini 3.1 Flash Lite: 7.8%

  • AA-LCRv1.1

    Reasoning across long documents

    • Qwen3 Max: 50.0%
    • Gemini 3.1 Flash Lite: 74.3%Top

  • MLCR-AA

    Long-context reasoning over medical records

    • Qwen3 Max: —
    • Gemini 3.1 Flash Lite: 5.0%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Qwen3 Max: 11.9%
    • Gemini 3.1 Flash Lite: 17.2%Top

  • GPQA Diamond

    Graduate-level science questions

    • Qwen3 Max: 76.4%
    • Gemini 3.1 Flash Lite: 82.2%Top

  • MMLU-Pro

    Multi-subject knowledge and reasoning

    • Qwen3 Max: 84.1%
    • Gemini 3.1 Flash Lite: —

  • IFBench

    Following precise output instructions

    • Qwen3 Max: 44.1%
    • Gemini 3.1 Flash Lite: 77.2%Top

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Qwen3 Max: —
    • Gemini 3.1 Flash Lite: 0.5%

  • SciCode

    Scientific computing in Python

    • Qwen3 Max: —
    • Gemini 3.1 Flash Lite: 43.4%

  • LiveCodeBench

    Recent competitive programming problems

    • Qwen3 Max: 76.7%
    • Gemini 3.1 Flash Lite: —

Math

  • AIME 2025

    Competition mathematics, 2025 problems

    • Qwen3 Max: 80.7%
    • Gemini 3.1 Flash Lite: —

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • Qwen3 Max: 20.5%
    • Gemini 3.1 Flash Lite: 24.2%Top

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Qwen3 Max: 74.3%Top
    • Gemini 3.1 Flash Lite: 31.3%

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Qwen3 Max: 50.0%
    • Gemini 3.1 Flash Lite: 74.3%Top

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-max",
    "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-max",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

Qwen

Other Qwen models, starting with the same family.