MiniMax-M2.5: Price, Context, Benchmarks, and Release Details
56.7
SI Score (method si-v2-absolute-shrinkage-1)
#68 of 115 ranked
88% confidence 88 percent, High confidence — 2 of 3 expected sources in
Coverage 70% of expected source weight · 100% confidence at 80% coverage
Pillar breakdown
Coding (weight 40 percent) 75.8
Math (weight 15 percent) —
Preference (weight 15 percent) 68.9
Reasoning (weight 30 percent) 33.3
Pillar weights: reasoning 30% · math 15% · coding 40% · preference 15%. Benchmark results use fixed 0–100 scales before averaging; incomplete evidence is shrunk toward 50.
Facts
- Input price / 1M
- not yet reported
- Output price / 1M
- not yet reported
- Context window
- 205Kmodels.devPublished source fact
Retrieved Oct 8, 2026 · MIT
Open source ↗ - Max output
- 131Kmodels.devPublished source fact
Retrieved Oct 8, 2026 · MIT
Open source ↗ - Released
- Feb 12, 2026models.devPublished source fact
Retrieved Oct 8, 2026 · MIT
Open source ↗ - Open weights
- Yesmodels.devPublished source fact
Retrieved Oct 8, 2026 · MIT
Open source ↗ - License
- not yet reported
- Input modalities
- textmodels.devPublished source fact
Retrieved Oct 8, 2026 · MIT
Open source ↗
Benchmark results
7 results| Benchmark | Raw result | Normalized (0–100) | Pillar |
|---|---|---|---|
| arc agi v1 public eval | 59.1%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] minimax-m2.5Published Oct 6, 2026
Retrieved Oct 8, 2026 · factual citation Open source ↗ | 59.1 | reasoning |
| arc agi v1 semi private | 63.7%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] minimax-m2.5Published Oct 6, 2026
Retrieved Oct 8, 2026 · factual citation Open source ↗ | 63.7 | reasoning |
| arc agi v2 public eval | 5.4%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] minimax-m2.5Published Oct 6, 2026
Retrieved Oct 8, 2026 · factual citation Open source ↗ | 5.4 | reasoning |
| arc agi v2 semi private | 4.9%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] minimax-m2.5Published Oct 6, 2026
Retrieved Oct 8, 2026 · factual citation Open source ↗ | 4.9 | reasoning |
| lmarena text | 1359.3 eloLMArena / ArenaPublished source fact [variant] text / overallPublished Oct 2, 2026
Retrieved Oct 8, 2026 · CC-BY-4.0 Open source ↗ | 68.9 | preference |
| swe bench verified | 75.8%Official model cards via models.devLab-reported; metric resolved; transcribed by MIT models.dev catalog; not independently evaluated [variant]
Retrieved Oct 8, 2026 · factual citation; MIT transcription Open source ↗ | 75.8 | coding |
| swe bench verified | 75.8%SWE-bench VerifiedSWE-bench published model plus agent result; harness retained, not a base model evaluation [variant] mini-SWE-agent; high; 2.0.0Published Feb 17, 2026
Retrieved Oct 8, 2026 · factual citation Open source ↗ | 75.8 | coding |
Normalization uses fixed absolute 0–100 scales for each unit, independently of other models that have a result on each benchmark. Coverage and evidence breadth still affect the composite; compare the evaluation conditions before reading a small score gap as decisive. “lab-reported” marks the provider's own published figure.
Sources: in and pending
in Reported (2)
- Official model cards via models.dev · arrived Oct 8, 2026
- LMArena / Arena · arrived Oct 8, 2026
pending Awaiting (1)
- LiveBench · carries 30% of expected weight
The confidence % rises as pending sources publish. Some sources never cover some models — that is why 100% confidence arrives at 80% of expected weight, not at full coverage.