Kimi K3 vs MiniMax-M3

At a glance

• Capability index

  • Kimi K3157.4
  • MiniMax-M3146.9

  Kimi K3 is 10.5 points higher.

• Price per 1M tokens

  • Kimi K3$6.00
  • MiniMax-M3$0.525

  MiniMax-M3 is 11 times cheaper.

• Context window

  • Kimi K31M tokens
  • MiniMax-M31M tokens

  They take in about the same.

• Released

  • Kimi K3Jul 16, 2026
  • MiniMax-M3Jun 1, 2026

  Kimi K3 is 45 days 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 Kimi K3 vs MiniMax-M3

Measure	Kimi K3Moonshot AI	MiniMax-M3MiniMax
Model
Maker	Moonshot AI	MiniMax
Sold here by	Moonshot AI	MiniMax (minimax.io)
API model ID	kimi-k3	MiniMax-M3
Released	Jul 16, 2026	Jun 1, 2026
Knowledge cutoff	Not reported	Jan 2025
Weights	Open: downloadable	Open: downloadable
Developer’s country	China	China
Hosts selling it	84including Moonshot AI directly	56including MiniMax directly
Training compute	2.0 × 10²⁵ FLOPEpoch AI estimate	Not published
Test results
Capability index	157.4 (best of these models)#15 of 274	146.9#68 of 274
LiveBench overall	79.2% (best of these models)#15 of 66	67.3%#61 of 66
Arena text	1488 (best of these models)#16 of 413	1440#96 of 413
LiveBench coding	81.5% (best of these models)#15 of 66	68.2%#64 of 66
Arena WebDev	1658 (best of these models)#13 of 138	1482#58 of 138
Arena coding	1541 (best of these models)#9 of 408	1494#86 of 408
LiveBench mathematics	84.4% (best of these models)#53 of 66	77.0%#65 of 66
Mock AIME 2024–2025	97.2% (best of these models)#31 of 297	71.1%#133 of 297
FrontierMath Tiers 1–3	72.2%#22 of 114	Not tested
FrontierMath Tier 4	39.0%#29 of 70	Not tested
Arena math	1501 (best of these models)#16 of 396	1432#102 of 396
SimpleQA Verified	50.6%#24 of 86	Not tested
LiveBench agentic coding	62.2% (best of these models)#10 of 66	40.7%#60 of 66
DeepSWE	68.5%#19 of 69	Not tested
LiveBench language	85.5% (best of these models)#11 of 66	76.8%#45 of 66
LiveBench instructions	71.4% (best of these models)#23 of 66	57.5%#61 of 66
Arena creative writing	1460 (best of these models)#26 of 411	1406#99 of 411
Arena instruction following	1488 (best of these models)#14 of 413	1434#88 of 413
LiveBench data analysis	78.7% (best of these models)#21 of 66	76.2%#34 of 66
LiveBench reasoning	90.7% (best of these models)#9 of 66	74.5%#61 of 66
GPQA Diamond	93.1% (best of these models)#20 of 319	90.9%#34 of 319
Chess puzzles	39.0% (best of these models)#31 of 227	14.0%#113 of 227
Arena hard prompts	1517 (best of these models)#9 of 413	1463#90 of 413
Price per million tokens
Input	$3.00	$0.30 (best of these models)
Output	$15.00	$1.20 (best of these models)
Blended, 3 input to 1 output	$6.00	$0.525 (best of these models)
Cached input	$0.30	$0.06 (best of these models)
Long requests	No separate price reported	$0.60 in, $2.40 outabove 512K tokens
The maker’s own price	This listing	This listing
Limits
Context window	1M tokens	1M tokens
Max input	1M tokens (best of these models)	512K tokens
Max output	1M tokens (best of these models)	512K tokens
Features
Reads	images, text and video	images, text and video
Produces	text	text
Reasoning	Yeseffort low, high, max; can be switched off; thinking budget can be set	Yeseffort low, medium, high; 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	Fixed by the provider	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.