MiniMax-M3 vs Qwen3.8 Max

At a glance

• Capability index

  • MiniMax-M3146.9
  • Qwen3.8 Max156.4

  Qwen3.8 Max is 9.5 points higher.

• Price per 1M tokens

  • MiniMax-M3$0.525
  • Qwen3.8 Max$3.00

  MiniMax-M3 is 5.7 times cheaper.

• Context window

  • MiniMax-M31M tokens
  • Qwen3.8 Max1M tokens

  They take in about the same.

• Released

  • MiniMax-M3Jun 1, 2026
  • Qwen3.8 MaxAug 3, 2026

  Qwen3.8 Max is 2 months newer.

Where each scores higher

On the 18 tests both have taken, each on its own scale. A gap of a point or two can sit within a test’s margin of error.

4 more tests have results for only one of them; the table below lists every result.

Everything side by side

Comparison of MiniMax-M3 vs Qwen3.8 Max

Measure	MiniMax-M3MiniMax	Qwen3.8 MaxAlibaba
Model
Maker	MiniMax	Alibaba
Sold here by	MiniMax (minimax.io)	Alibaba
API model ID	MiniMax-M3	qwen3.8-max
Released	Jun 1, 2026	Aug 3, 2026
Knowledge cutoff	Jan 2025	Not reported
Weights	Open: downloadable	Closed: hosted access only
Developer’s country	China	China
Hosts selling it	56including MiniMax directly	36including Alibaba directly
Test results
Capability index	146.9#68 of 274	156.4 (best of these models)#22 of 274
LiveBench overall	67.3%#61 of 66	78.5% (best of these models)#17 of 66
Arena text	1440#96 of 413	1482 (best of these models)#23 of 413
LiveBench coding	68.2%#64 of 66	72.9% (best of these models)#52 of 66
Arena WebDev	1482#58 of 138	1671 (best of these models)#10 of 138
Arena coding	1494#86 of 408	1524 (best of these models)#28 of 408
LiveBench mathematics	77.0%#65 of 66	91.3% (best of these models)#27 of 66
Mock AIME 2024–2025	71.1%#133 of 297	99.4% (best of these models)#14 of 297
Arena math	1432#102 of 396	1498 (best of these models)#18 of 396
FrontierMath Tiers 1–3	Not tested	74.7%#18 of 114
FrontierMath Tier 4	Not tested	46.3%#25 of 70
SimpleQA Verified	Not tested	45.8%#37 of 86
LiveBench agentic coding	40.7%#60 of 66	64.7% (best of these models)#7 of 66
DeepSWE	Not tested	57.5%#36 of 69
LiveBench language	76.8%#45 of 66	79.7% (best of these models)#34 of 66
LiveBench instructions	57.5%#61 of 66	74.1% (best of these models)#13 of 66
Arena creative writing	1406#99 of 411	1468 (best of these models)#20 of 411
Arena instruction following	1434#88 of 413	1474 (best of these models)#31 of 413
LiveBench data analysis	76.2%#34 of 66	78.4% (best of these models)#24 of 66
LiveBench reasoning	74.5%#61 of 66	88.2% (best of these models)#23 of 66
GPQA Diamond	90.9%#34 of 319	92.7% (best of these models)#24 of 319
Chess puzzles	14.0%#113 of 227	29.0% (best of these models)#54 of 227
Arena hard prompts	1463#90 of 413	1503 (best of these models)#27 of 413
Price per million tokens
Input	$0.30 (best of these models)	$2.00
Output	$1.20 (best of these models)	$6.00
Blended, 3 input to 1 output	$0.525 (best of these models)	$3.00
Cached input	$0.06 (best of these models)	$0.25
Writing to the cache	Not reported	$2.50
Long requests	$0.60 in, $2.40 outabove 512K tokens	No separate price reported
The maker’s own price	This listing	This listing
Limits
Context window	1M tokens	1M tokens
Max input	512K tokens	991K tokens (best of these models)
Max output	512K tokens (best of these models)	131.1K tokens
Features
Reads	images, text and video	images, PDFs, text and video
Produces	text	text
Reasoning	Yeseffort low, medium, high; can be switched off; thinking budget can be set	Yeseffort low, medium, xhigh; can be switched off; thinking budget can be set
Tool calling	Yes	Yes
Structured output	Yes, such as JSON	Yes, such as JSON
File attachments	Yes	Yes
Temperature setting	Supported	Supported

Bold marks the better value in each row. Prices are each model’s own list price, or the middle price across hosts when the maker does not sell it, blended as three input tokens for every output token. Test results as published by LiveBench (Apache-2.0), Epoch AI (CC BY 4.0), Datacurve DeepSWE, compiled by Epoch AI (CC BY 4.0), Arena (CC BY 4.0); model details and prices from models.dev (MIT). Confirm pricing with the provider before use.