MiniMax M2.7
MiniMax's March 2026 open-weight agentic coding model, since succeeded by MiniMax M3 (June 2026).
Our editorial assessment, based on the public sources listed below. Last reviewed .
- Overall rank
- #47
- Tier
- A
- Score
- 79.2
- Main category
- Coding
Strengths
- MiniMax reports SWE-Pro results approaching Anthropic's Opus models
- MiniMax reports strong GDPval-AA results among open-source models
- Efficient architecture with 10 billion active parameters
Considerations
- Self-hosting the open weights requires your own infrastructure
- Succeeded by MiniMax M3 in June 2026
Price
$0.30 / 1M tokens
Shown for reference — price is not part of the score.
Score breakdown
- Performance
- 82.0
- 55% of score
- Reputation
- 78.0
- 35% of score
- Recency
- 68.0
- 10% of score
The overall score combines these sub-scores using the weights shown. How we score →
Verification status
Each field carries its own re-verification date. Amber means older than 90 days — treat those figures as stale until re-checked.
- Pricing
- Benchmark
- Reputation
- Feature list
Checked against the sources listed below.
Sources
- Vals AI — MiniMax-M2.7 benchmark results (opens in a new tab)Accessed:
- MarkTechPost — MiniMax open-sources M2.7 (opens in a new tab)Accessed:
- MiniMax M2.7 announcement (opens in a new tab)Accessed:
- MarkTechPost: MiniMax releases M3 (June 2026 successor) (opens in a new tab)Accessed:
- TeamDay.ai - Frontier AI Models 2026 (opens in a new tab)Accessed:
Show 1 more source
Similar models
- Claude Opus 4.7Anthropic#35 · Tier A · Score 82.0Compare
- Claude Sonnet 4.6Anthropic#40 · Tier A · Score 81.3Compare
- Kimi K2.7 CodeMoonshot AI#53 · Tier A · Score 76.9Compare
- Kimi K2.6Moonshot AI#55 · Tier A · Score 75.8Compare