MiniMax M3

by MiniMax AI

MiniMax M3 is a frontier-tier open-weight multimodal agent model with 428B total parameters (23B activated) and a 1M token context window. Released May 31, 2026, it features MiniMax Sparse Attention (MSA) for 9x prefill and 15x decode speedup over its predecessor, and is specialized for coding, agentic tasks, and complex reasoning.

Multimodal Models
Proprietary

Primary Use Cases

  • Long-context document analysis and reasoning
  • Agentic software engineering tasks
  • Multimodal content understanding (text, image, video)
  • Complex coding and debugging

✅ Strengths

  • Highly competitive pricing ($0.30/$1.20 per 1M tokens at promotional rate)
  • 1M token context window enables very long document processing
  • Open weights available on Hugging Face for self-hosting
  • Strong benchmark performance on video understanding and coding tasks

⚠️ Limitations

  • Promotional pricing may increase after introductory period
  • Long-context inputs (>512K tokens) incur 2x pricing
  • Agent-oriented use cases may incur additional costs from tool-calling and retries
  • Less established ecosystem compared to OpenAI or Anthropic

🏆 Performance Benchmarks

Standardized benchmark scores for objective comparison

MME-Benchmarks/Video(Multimodal Video Understanding)
0rank

Ranked #1 on MME-Benchmarks/Video as of June 2026

Technical Specifications

Developer
MiniMax AI
Category
Multimodal Models
Type
🧠 Foundation Model
License
UNKNOWN

Performance Benchmarks

MME-Benchmarks/Video0rank

Source: View benchmark details

Trust & Privacy

Health Status
🟢 Active
Trains on User Data
❓ Unknown
Certifications
None listed

Ecosystem & Integrations

API Access
✅ Available
Chrome Extension
❌ No
Mobile App
❌ No
Pricing Model
💰 Pay Per Use

Pricing Breakdown

Standard (≤512K context)

Promotional pricing for inputs up to 512K tokens

Input$0.3/M tokens
Output$1.2/M tokens
0
Long-Context (>512K)

For inputs exceeding 512K tokens

Input$0.6/M tokens
Output$2.4/M tokens
0