LIVE
Policy11/2OpenAI launches Grid Protectors Program to help protect US data center power grids
More ›Submit for Promotion
← Back to model catalog

MiniMax

MiniMax M3

BOpen

A new-generation flagship model released and open-sourced by MiniMax on June 1, 2026, featuring a self-developed MSA sparse attention architecture (not MoE), 1M context, and native multimodality. SWE-Bench Pro 59.0% surpasses GPT-5.5, and BrowseComp 83.5 points surpasses Claude Opus 4.7. One week after release, it ranked third globally in OpenRouter call volume.

API Pricing

Input

$0.3

Output

$1.2

Cached input

$0.06

Peak hours

—

USD / 1M tokens

※ Overseas pricing (MiniMax official pricing page, checked 2026-10-20): standard tier $0.30 input / $1.20 output / $0.06 prompt-cache read per 1M tokens for ≤512k input tokens; >512k input tier $0.60/$2.40; priority tier 1.5x standard. Mainland-China pricing is shown on the Chinese page

Price sourceplatform.minimax.io ↗cloud.tencent.com ↗· verified 2026-10-20

Latest change：Up 140% vs previous（$0.294 → $0.706 / 1M）

Price history

12 snapshots · 2026-04-28 → 2026-08-17

Current: $0.706Period low: $0.294 (2026-04-28)Period high: $0.706 (2026-08-17)First tracked: $0.294 (2026-04-28)

Specifications

Context window
1.05M Tokens
Max output
512K Tokens
Total parameters
427B
Architecture
MSA sparse attention
Input modalities
Text · Image · Video
License
Apache 2.0

Overall score

Total

51.6

Rank

#19

Tier

B

Data confidence

85%

Capability34
Scale35
Value79
Context80.8
Ecosystem77.5
Freshness26.7

Capabilities & use cases

SWE-Bench_Pro 59.0%BrowseComp 83.5

Key capabilities

  • Native multimodal
  • Desktop computer control
  • Prefill 9.7x speedup
  • Decoding 15.6x speedup

Best for

  • AI Coding (repository-level code analysis)
  • Multimodal Agent
  • Long document/video understanding
  • Browser automation
  • Multi-step tool calling

Official links

Website ↗API platform ↗Pricing page ↗Open in comparator ↗

Same vendor

MiniMax M3.1 Flash

Similar price

GPT-5.6 Luna $1.2Qwen3.7-Plus (Qwen) $1.6DeepSeek-V4.1-Flash $0.6GLM-5.3-Flash $0.5Claude Haiku 5.5 $0.5GPT-6 Luna $0.5

FAQ

What does MiniMax M3 cost?
MiniMax M3 is priced at $1.2 output and $0.3 input (USD per 1M tokens)
How large is the context window of MiniMax M3?
MiniMax M3 has a 1.05M Tokens context window with up to 512K Tokens output.
Where can I access MiniMax M3?
Official channels: Website and API platform.

Data is refreshed daily and cross-checked against two independent sources; please confirm pricing and specs on the vendor page.