MiniMax-M3
| Released | 2026-06-01 |
|---|---|
| Price per 1M | $0.30 in · $1.20 out |
| Overall score | 53.6 · rank 84 |
| Context window | 1,048,576 tokens |
| Provider | MiniMax |
| Weights | Open weights |
Scores
| Dimension | Score | Rank | Benchmarks |
|---|---|---|---|
| Overall: Balanced | 53.6 | 84 | 36 |
| Overall: Coding | 53.2 | 84 | 26 |
| Overall: Reasoning | 53.3 | 88 | 10 |
| Overall: Agentic work | 54.4 | 81 | 24 |
| Overall: Frontend | 52.1 | 75 | 11 |
| Coding | 52.8 | 96 | 7 |
| Agents | 55.0 | 74 | 11 |
| Reasoning | 54.7 | 85 | 6 |
| Math | 52.4 | 93 | 2 |
| Knowledge | 51.6 | 85 | 2 |
| Frontend | 51.6 | 69 | 2 |
| General | 55.9 | 61 | 4 |
| Multimodal | 53.1 | 60 | 2 |
| Value vs price | +7.8 | 47 |
Configurations
| Configuration | Balanced | Benchmarks |
|---|---|---|
| Unspecified setting | 54.0 ±0.9 | 35 |
| Non-reasoning | 47.4 ±3.8 | 5 |
Evidence
Coding 52.8 ±2.2
| Benchmark | Result | Source | Date |
|---|---|---|---|
| FrontierCode | 14.7% | Epoch AI | 2026-11-11 |
| Code Migration v1.1 | 20.9% | Vals AI | 2026-11-09 |
| SWE-rebench | 47.2% | SWE-rebench | 2026-07-01 |
| SWE-bench Multimodal | 33.3% | Vals AI | 2026-11-09 |
| SWE-bench Verified (Vals run) | 75.0% | Vals AI | 2026-09-01 |
| SciCode | 47.1% | Epoch AI | 2026-11-11 |
| LiveCodeBench (Vals run) | 82.2% | Vals AI | 2026-09-01 |
Agents 55.0 ±1.0
| Benchmark | Result | Source | Date |
|---|---|---|---|
| Terminal-Bench 4.0 (Vals run) | 1.0% | Vals AI | 2026-11-09 |
| OSWorld 2.0 | 4.6% | Epoch AI | 2026-11-11 |
| APEX-Agents | 37.7% | Epoch AI | 2026-11-11 |
| Terminal-Bench 2.1 (Vals run) | 53.6% | Vals AI | 2026-10-01 |
| OSWorld-Verified | 75.2% | OSWorld-Verified | 2026-06-08 |
| EMB v1.1 (Excel financial modeling) | 59.8% | Vals AI | 2026-11-09 |
| Tax Agent Bench | 20.6% | Vals AI | 2026-11-09 |
| Finance Agent v2.1 | 54.3% | Vals AI | 2026-11-09 |
| SkillsBench | 51.5% | Vals AI | 2026-10-29 |
| Vending-Bench 2 | $2,158 | Epoch AI | 2026-11-11 |
| CyberBench v1.1 | 65.0% | Vals AI | 2026-11-05 |
Reasoning 54.7 ±3.5
| Benchmark | Result | Source | Date |
|---|---|---|---|
| Extended NYT Connections | 65.1% | Lech Mazur | 2026-11-10 |
| CritPt | 3.7% | Epoch AI | 2026-11-11 |
| GPQA Diamond | 90.9% | Epoch AI | 2026-08-10 |
| GPQA Diamond (Vals run) | 92.7% | Vals AI | 2026-09-01 |
| Chess Puzzles | 17.0% | Epoch AI | 2026-10-13 |
| SimpleBench | 45.8% | SimpleBench | 2026-07-09 |
| SimpleBench | 45.8% | Epoch AI | 2026-11-11 |
| Mystery Game Puzzles | 8.0% | Epoch AI | 2026-08-27 |
Math 52.4 ±3.4
| Benchmark | Result | Source | Date |
|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | Epoch AI | 2026-08-10 |
| ProofBench v1.1 | 18.0% | Vals AI | 2026-11-09 |
| ProofBench | 18.0% | Epoch AI | 2026-11-11 |
Knowledge 51.6 ±3.9
| Benchmark | Result | Source | Date |
|---|---|---|---|
| MMLU-Pro (Vals run) | 84.2% | Vals AI | 2026-09-01 |
| Legal Research Bench | 29.8% | Vals AI | 2026-11-09 |
Frontend 51.6 ±0.7
| Benchmark | Result | Source | Date |
|---|---|---|---|
| Vibe Code Bench v1.1 | 47.6% | Vals AI | 2026-11-09 |
| LMArena WebDev | 1,484 | LMArena | 2026-11-09 |
General 55.9 ±1.9
| Benchmark | Result | Source | Date |
|---|---|---|---|
| LLM Debate Benchmark | 1,496 | Lech Mazur | 2026-10-26 |
| EQ-Bench 4 | 1,140 | EQ-Bench | 2026-10-24 |
| LMArena Text | 1,436 | LMArena | 2026-11-09 |
| Creative Story-Writing (Lech Mazur) | 48.0% | Lech Mazur | 2026-10-23 |
Multimodal 53.1 ±2.3
| Benchmark | Result | Source | Date |
|---|---|---|---|
| MMMU-Pro (Vals run) | 81.2% | Vals AI | 2026-09-01 |
| LMArena Vision | 1,232 | LMArena | 2026-11-09 |
Results not used in the score
| Benchmark | Result | Source | Why not used |
|---|---|---|---|
| Chess Puzzles | 8.0% | Epoch AI | A stronger configuration result was used |
| GPQA Diamond | 81.3% | Epoch AI | A stronger configuration result was used |
| Mystery Game Puzzles | 7.0% | Epoch AI | A stronger configuration result was used |
| OTIS Mock AIME 2024-2025 | 26.7% | Epoch AI | A stronger configuration result was used |
Names matched to this model
minimax-m3