MiniMax-M3
- Input
- $0.30/M tok
- Output
- $1.20/M tok
- Blended
- $0.525/M tok · 3:1
- Speed
- 88tok/s
- Context
- 1Mtokens
- Cost per task
- ~$0.51/taskest.
Open weights (minimax-community license). Price includes a permanent 50% discount.
Rankings
Coding
Mixed#64 / 137rank by quality46.1±5.4quality / 1004/7 benchmarks measuredWriting & chat
Verified#31 / 162rank by quality68.7±4.1quality / 1003/5 benchmarks measuredVision
Mixed#35 / 98rank by quality63.2±5.4quality / 1003/6 benchmarks measuredHard reasoning
Verified#52 / 218rank by quality66.0±3.7quality / 1003/4 benchmarks measuredAgents
Mixed#47 / 187rank by quality46.6±3.4quality / 1005/7 benchmarks measured
Self-hostingEstimated
How we estimate →What it would cost to run these open weights yourself on rented GPUs, compared with the API price above.
- Hardware
- 8× H200 141GB
- FP8 weights
- Throughput
- ~20,773 tok/s
- many requests batched
- Price
- $0.18–1.46 /M tok
- busy → light use
- On your own machine
- Mac Studio M5 Ultra 512GB (int4, ~73 tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- FP8 weights (428B params, 23B active per token) + 40% KV-cache headroom ≈ 599 GB
- 8× H200 141GB at $3.59–$7.91/GPU-hour on-demand (2026-09-24)
- ~20,773 output tok/s aggregate at batch 747 (bandwidth-bound); MoE compute scales with active params
- Blended 3:1 input:output; prefill ~162,622 tok/s
- Low = 75% utilization at the low GPU price; high = 20% at the high price
Benchmark scores
~ italic, dashed = no published score yet, estimated from related benchmarks.
Coding
DeepSWE v1.1Estimated
~#38 of 46 (estimated)~49.5%
- #1Gemini 4 Argon78.8%
- #2DeepSeek V4.1 Flash 101777.0%
- #37Qwen3.8 Max (0902)51.0%
- ~#38MiniMax-M3Estimated49.5%
- #38Ripple-65B Preview44.3%
No published score — estimated from related benchmarks
Terminal-Bench 4.0Verified
#43 of 1012.0%
- #1Kimi K3.1 Pro67.2%
- #2Claude Sonnet 5.563.6%
- #43GPT-5.4 mini2.0%
- #43MiniMax-M32.0%
- #45K2 Horizon 375B A23B1.5%
artificialanalysis.ai · 2026-09-24
LiveCodeBenchEstimated
~#15 of 28 (estimated)~73.3%
- #1Qwen3.8-Omni-Flash92.6%
- #2Qwen3.8 Flash-Next91.9%
- #14Maple-Preview75.1%
- ~#15MiniMax-M3Estimated73.3%
- #15Mistral Small 464.0%
No published score — estimated from related benchmarks
AA Coding Agent IndexEstimated
~#23 of 26 (estimated)~36.0
- #1Claude Opus 5.561.7
- #2Claude Opus 559.7
- #22Claude Haiku 5.537.3
- ~#23MiniMax-M3Estimated36.0
- #23Gemini 3.8 Flash33.3
No published score — estimated from related benchmarks
Writing & chat
LMArena Text, Non-English (Elo)Verified
#33 of 561427 Elo
- #1Gemini 4 ArgonFew votes1512
- #2Claude Opus 5.51497
- #32Qwen3.8 27B1429
- #33MiniMax-M31427
- #34Gemma 4 26B A4BFew votes1423
lmarena.ai · 2026-09-24
EQ-Bench Creative Writing v3 (Elo)Estimated
~#10 of 26 (estimated)~1865 Elo
- #1GPT-6 Astra2173
- #2Claude Fable 5.12162
- #9Muse Spark 1.31906
- ~#10MiniMax-M3Estimated1865
- #10GPT-5.6 Terra1855
No published score — estimated from related benchmarks
Vision
MMMU-ProVerified
#22 of 8778.6%
- #1Claude Opus 5.587.7%
- #2GPT-6 Astra86.9%
- #20Ling-3.0-flash-VL79.0%
- #22MiniMax-M378.6%
- #23GPT-5.3 Codex78.5%
artificialanalysis.ai · 2026-09-24
CharXiv ReasoningEstimated
~#8 of 12 (estimated)~79.6%
- #1Gemini 3.8 Flash86.2%
- #2Kimi K384.8%
- #7Nemotron 3.5 Super VL NVFP479.9%
- ~#8MiniMax-M3Estimated79.6%
- #8Muse Glimmer78.8%
No published score — estimated from related benchmarks
GDP.pdf (AA)Verified
#34 of 479.8%
- #1Muse Wonder Full32.6%
- #2GPT-6 Astra31.0%
- #33Muse Glimmer10.0%
- #34MiniMax-M39.8%
- #35Ling-3.0-flash-VL8.4%
artificialanalysis.ai · 2026-09-24
ChartographyEstimated
~#16 of 25 (estimated)~20.7%
- #1Claude Fable 5.273.4%
- #2GPT-6 Astra71.0%
- #15Gemini 3.1 Pro26.1%
- ~#16MiniMax-M3Estimated20.7%
- #16Grok 4.620.5%
No published score — estimated from related benchmarks
Hard reasoning
GPQA DiamondVerified
#12 of 19292.9%
- #1GPT-6 Astra96.1%
- #2Gemini 3.8 Flash95.3%
- #11Claude Opus 593.2%
- #12MiniMax-M392.9%
- #13DeepSeek V4 Pro (0813)92.8%
artificialanalysis.ai · 2026-09-24
Humanity's Last ExamVerified
#37 of 19839.0%
- #1Claude Fable 5.274.3%
- #2Claude Opus 5.561.4%
- #36DeepSeek V4.1 Flash39.2%
- #37MiniMax-M339.0%
- #38GPT-6 Luna38.5%
artificialanalysis.ai · 2026-09-24
AIME (latest)Estimated
~#8 of 34 (estimated)~94.2%
- #1GLM-5.3 NVFP499.5%
- #2Beam 1.0 501B A23B97.7%
- #7Nova 2.0 Lite94.3%
- ~#8MiniMax-M3Estimated94.2%
- #8gpt-oss-120b93.4%
No published score — estimated from related benchmarks
Agents
τ²-bench (Telecom)Verified
#18 of 11988.9%
- #1Step 3.7 Flash98.5%
- #2DeepSeek-V4-Pro-0813-nvfp4-DSpark98.3%
- #17KAT Coder Pro V289.5%
- #18MiniMax-M388.9%
- #19Qwen3.5 Omni Plus88.3%
artificialanalysis.ai · 2026-09-24
OSWorld-VerifiedEstimated
~#11 of 15 (estimated)~69.1%
- #1Kimi K384.8%
- #2Qwen3.8 27B84.3%
- #10Nex-N2.5-Mini71.2%
- ~#11MiniMax-M3Estimated69.1%
- #11Muse Glimmer65.9%
No published score — estimated from related benchmarks
OSWorld 2.0Estimated
~last of 5 (estimated)below measured range (<47.5%)
No published score — estimated from related benchmarks
AutomationBench-AAVerified
#39 of 8421.3%
- #1Claude Fable 5.276.4%
- #2Muse Wonder Full71.3%
- #38Quasar 438B23.9%
- #39MiniMax-M321.3%
- #40K2 Horizon 7B20.9%
artificialanalysis.ai · 2026-09-24
Not comparable across labs (1)
Kept for reference, not counted in rankings: these use a lab-specific task set, answer key or scoring, so scores can't be compared fairly across labs. Only published scores are shown.