MiMo-V2.6-Flash
- Input
- $0.14/M tok
- Output
- $0.28/M tok
- Blended
- $0.175/M tok · 3:1
- Speed
- 55tok/s
- Context
- 1Mtokens
- Cost per task
- ~$0.12/taskest.
Open weights (MIT); 309B total / 15B active omnimodal MoE. Not yet on Artificial Analysis (no speed).
Rankings
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
- 4× B200 180GB
- FP8 weights
- Throughput
- ~18,660 tok/s
- many requests batched
- Price
- $0.14–1.27 /M tok
- busy → light use
- On your own machine
- Mac Studio M5 Ultra 512GB (int4, ~110 tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- FP8 weights (309B params, 15B active per token) + 40% KV-cache headroom ≈ 433 GB
- 4× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
- ~18,660 output tok/s aggregate at batch 534 (bandwidth-bound); MoE compute scales with active params
- Blended 3:1 input:output; prefill ~283,500 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
SWE-bench VerifiedEstimated
~#3 of 21 (estimated)~85.0%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Estimated
~#7 of 24 (estimated)~67.4%
- #1Claude Fable 5.293.0%
- #2Claude Opus 5.589.9%
- #6Intern-S2-397B68.5%
- ~#7MiMo-V2.6-FlashEstimated67.4%
- #7Beam 1.0 501B A23B66.1%
No published score — estimated from related benchmarks
DeepSWE v1.1Lab-reported
#17 of 4567.9%
- #1Gemini 4 Argon78.8%
- #2DeepSeek V4.1 Flash 101777.0%
- #15Kimi K368.4%
- #17MiMo-V2.6-Flash67.9%
- #18GPT-6 Astra67.6%
huggingface.co · 2026-09-21
Terminal-Bench 4.0Verified
#29 of 10122.7%
- #1Kimi K3.1 Pro67.2%
- #2Claude Sonnet 5.563.6%
- #28Qwen3.8 Flash-Next25.3%
- #29MiMo-V2.6-Flash22.7%
- #30SWE-2 Flash20.0%
artificialanalysis.ai · 2026-09-28
LiveCodeBenchEstimated
~#10 of 28 (estimated)~86.7%
- #1Qwen3.8-Omni-Flash92.6%
- #2Qwen3.8 Flash-Next91.9%
- #9Nemotron 3.5 Super VL NVFP487.4%
- ~#10MiMo-V2.6-FlashEstimated86.7%
- #10Ternary Bonsai 27B82.8%
No published score — estimated from related benchmarks
SciCodeVerified
#42 of 11851.3%
- #1Claude Fable 5.267.7%
- #2Claude Opus 5.566.9%
- #41GLM-5.3 Flash51.6%
- #42MiMo-V2.6-Flash51.3%
- #43DeepSeek V4 Pro (0813)51.0%
artificialanalysis.ai · 2026-09-28
AA Coding Agent IndexEstimated
~#20 of 26 (estimated)~42.6
- #1Claude Opus 5.561.7
- #2Claude Opus 559.7
- #18Step 5 Preview43.1
- ~#20MiMo-V2.6-FlashEstimated42.6
- #20Kimi K341.4
No published score — estimated from related benchmarks
Agents
τ²-bench (Telecom)Estimated
~#3 of 120 (estimated)~96.8%
- #1Step 3.7 Flash98.5%
- #2DeepSeek-V4-Pro-0813-nvfp4-DSpark98.3%
- ~#3MiMo-V2.6-FlashEstimated96.8%
- #3Gemini 3.1 Pro95.6%
No published score — estimated from related benchmarks
OSWorld 2.0Estimated
~#3 of 5 (estimated)~53.8%
No published score — estimated from related benchmarks
BrowseCompEstimated
~#8 of 22 (estimated)~86.4%
- #1Atria Dawn Preview92.5%
- #2GPT-6 Astra91.5%
- #7GPT-5.6 Terra87.5%
- ~#8MiMo-V2.6-FlashEstimated86.4%
- #8Gemini 3.1 Pro85.9%
No published score — estimated from related benchmarks
GDPval-AA (Elo)Estimated
~#20 of 92 (estimated)~1568 Elo
- #1Claude Opus 5.51866
- #2Muse Wonder Full1857
- #19GPT-6 Astra1574
- ~#20MiMo-V2.6-FlashEstimated1568
- #20Kimi K31533
No published score — estimated from related benchmarks
AutomationBench-AAEstimated
~#27 of 85 (estimated)~48.5%
- #1Claude Fable 5.276.4%
- #2Muse Wonder Full71.3%
- #26Step 5 Preview51.0%
- ~#27MiMo-V2.6-FlashEstimated48.5%
- #27Qwen3.8 27B48.2%
No published score — estimated from related benchmarks
τ-Bench Banking (AA)Estimated
~#18 of 91 (estimated)~37.0%
- #1Muse Spark 1.350.5%
- #2GLM-5.350.3%
- #17Claude Sonnet 537.3%
- ~#18MiMo-V2.6-FlashEstimated37.0%
- #18Motif 335.3%
No published score — estimated from related benchmarks