Qwen: Qwen3 Max Thinking

Qwen: Qwen3 Max Thinking

qwen · Released Feb 9, 2026
67
Our Score

Qwen3-Max-Thinking is the flagship reasoning model in the Qwen3 series, designed for high-stakes cognitive tasks that require deep, multi-step reasoning. By significantly scaling model capacity and reinforcement learning compute, it delivers major gains in factual accuracy, complex reasoning, instruction following, alignment with human preferences, and agentic behavior.

$0.78 / 1M Input Price
$3.90 / 1M Output Price
262,144 tokens Context Window
32,768 tokens Max Output

Capabilities

Tool Use Function Calling

Architecture

ModalityText → Text
TokenizerQwen

Performance Indices

Source: Artificial Analysis

32.5 Intelligence Index
24.5 Coding Index
50.5 Agentic Index
82.3 Math Index

Benchmark Scores

Evaluations

GPQA Diamond 77.6%
Graduate-level scientific reasoning
HLE 12%
Humanity's Last Exam
MMLU Pro 82.4%
Multi-task language understanding
LiveCodeBench 53.5%
Live coding evaluation
SciCode 38.7%
Scientific computing
AIME 2025 82.3%
Competition mathematics (2025)
IFBench 53.8%
Instruction following
LCR 57.7%
Long-context reasoning
TerminalBench Hard 17.4%
Agentic terminal tasks
τ²-Bench 83.6%
Conversational agent benchmark

Benchmark data from Artificial Analysis and Hugging Face

Model Information

OpenRouter ID qwen/qwen3-max-thinking
Providerqwen
Release Date February 9, 2026
Context Length262,144 tokens
Max Completion32,768 tokens
Status Active

Pricing

Token Type Cost per 1M tokens Cost per 1K tokens
Input $0.78 $0.000780
Output $3.90 $0.003900

Live Performance

Live endpoint metrics — refreshed every 30 minutes.

100%
Avg Uptime
1,308ms
Best Latency (TTFT)
30 tok/s
Best Throughput
1/1
Active Endpoints
Available via: Alibaba

Leaderboard Categories