qwen

Qwen: Qwen3 Max Thinking

Qwen3 Max Thinking is a text-focused AI model capable of handling extensive context up to 262,144 tokens and supporting both reasoning and tools integration for enhanced problem-solving. This model excels in understanding complex texts without needing structured output specifically. Its benchmark score of 52.2 across independent evaluations indicates strong performance, though the coverage is complete. For users prioritizing cost-effectiveness, Qwen3 Max Thinking stands out with a pricing structure of $0.78 per million input tokens and $3.9 per million output tokens, making it competitive in scenarios requiring extensive text processing. Its blended benchmark score of 52.2 places it among top performers, ensuring reliable performance across various tasks.

Quality Score
97/100
price + capability + benchmarks
Input Price
$0.78
per 1M tokens
Output Price
$3.90
per 1M tokens
Context Window
262,144
tokens

Benchmark results

Independent, published benchmarks. Blended score 51.5 across 1 benchmark, last refreshed 2026-07-27. How scoring works →

BenchmarkMeasuresScore
AI Index broad capability composite 52.4
Model ID
qwen/qwen3-max-thinking
Vendor
qwen
Released
February 2026
Tokenizer
Qwen
Input Modalities
text
Output Modalities
text
Max Output
32,768 tokens
Tool Calling
✓ supported
Structured Output
✓ supported
Reasoning Mode
✓ supported
Vision
text only
Audio
no
Moderated
no

What it costs in practice

Computed from the current $0.78/M input and $3.90/M output rates. Run your own numbers →

JobTokensCost
Summarize a 50-page report 30k in / 1.5k out $0.03
Classify 1,000 customer emails 500k in / 50k out $0.58
A month of a busy support chatbot 5M in / 2M out $11.70

Similar models

Quick answers

How much does Qwen: Qwen3 Max Thinking cost?
$0.78 per million input tokens and $3.90 per million output tokens.
What is Qwen: Qwen3 Max Thinking's context window?
262,144 tokens, roughly 393 pages of text in a single request.
Does Qwen: Qwen3 Max Thinking support tool calling?
It supports tool calling, structured output, a reasoning mode.
Can Qwen: Qwen3 Max Thinking process images?
No, it is text-only on the input side.