Step 3.7 Flash
- Input
- $0.20/M tok
- Output
- $1.15/M tok
- Blended
- $0.438/M tok · 3:1
- Speed
- 137tok/s
- Context
- 262Ktokens
- Cost per task
- ~$0.48/taskest.
Long-tail import from Artificial Analysis (2026-09-24); AA variant: Step 3.7 Flash. Price = AA median host price.
Rankings
Coding
Verified#67 / 137rank by quality43.8±10.1quality / 1001/7 benchmarks measuredWriting & chat
Verified#54 / 162rank by quality60.2±9.5quality / 1001/5 benchmarks measuredVision
Verified#41 / 98rank by quality60.3±8.2quality / 1001/6 benchmarks measuredHard reasoning
Verified#88 / 218rank by quality51.7±5.9quality / 1002/4 benchmarks measuredAgents
Verified#66 / 187rank by quality41.2±5.9quality / 1003/7 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
- 2× B200 180GB
- FP8 weights
- Throughput
- ~8,082 tok/s
- many requests batched
- Price
- $0.15–1.38 /M tok
- busy → light use
- On your own machine
- Mac Studio M5 Max 128GB (int4, ~78 tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- FP8 weights (198B params, 11B active per token) + 25% KV-cache headroom ≈ 248 GB
- 2× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
- ~8,082 output tok/s aggregate at batch 231 (bandwidth-bound); MoE compute scales with active params
- Blended 3:1 input:output; prefill ~193,295 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
~#13 of 21 (estimated)~73.4%
- #1Claude Opus 596.0%
- #2Claude Sonnet 585.2%
- #12Flint74.2%
- ~#13Step 3.7 FlashEstimated73.4%
- #13Claude Haiku 4.573.3%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Estimated
~#20 of 24 (estimated)~53.2%
- #1Claude Fable 5.293.0%
- #2Claude Opus 5.589.9%
- #18Gemini 3.5 Flash-Lite54.2%
- ~#20Step 3.7 FlashEstimated53.2%
- #20Nemotron 3.5 Super VL51.8%
No published score — estimated from related benchmarks
DeepSWE v1.1Estimated
~#38 of 46 (estimated)~45.7%
- #1Gemini 4 Argon78.8%
- #2DeepSeek V4.1 Flash 101777.0%
- #37Qwen3.8 Max (0902)51.0%
- ~#38Step 3.7 FlashEstimated45.7%
- #38Ripple-65B Preview44.3%
No published score — estimated from related benchmarks
Terminal-Bench 4.0Estimated
~#36 of 102 (estimated)~13.6%
- #1Kimi K3.1 Pro67.2%
- #2Claude Sonnet 5.563.6%
- #34DeepSeek V4 Pro (0813)14.1%
- ~#36Step 3.7 FlashEstimated13.6%
- #36GPT-5.5 Instant (June 2026)12.6%
No published score — estimated from related benchmarks
LiveCodeBenchEstimated
~#15 of 28 (estimated)~69.9%
- #1Qwen3.8-Omni-Flash92.6%
- #2Qwen3.8 Flash-Next91.9%
- #14Maple-Preview75.1%
- ~#15Step 3.7 FlashEstimated69.9%
- #15Mistral Small 464.0%
No published score — estimated from related benchmarks
SciCodeVerified
#66 of 11843.9%
- #1Claude Fable 5.267.7%
- #2Claude Opus 5.566.9%
- #66MiMo-V2.543.9%
- #66Step 3.7 Flash43.9%
- #68Nex-N2-Pro43.3%
artificialanalysis.ai · 2026-09-24
AA Coding Agent IndexEstimated
~#23 of 26 (estimated)~34.3
- #1Claude Opus 5.561.7
- #2Claude Opus 559.7
- #22Claude Haiku 5.537.3
- ~#23Step 3.7 FlashEstimated34.3
- #23Gemini 3.8 Flash33.3
No published score — estimated from related benchmarks
Writing & chat
IFBenchVerified
#42 of 13967.3%
- #1MiniMax-M382.9%
- #2Qwen3.8 2.4T A95B82.8%
- #41Mistral Medium 3.568.8%
- #42Step 3.7 Flash67.3%
- #43MiMo-V2.567.1%
artificialanalysis.ai · 2026-09-24
LMArena Text, Non-English (Elo)Estimated
~#36 of 57 (estimated)~1417 Elo
- #1Gemini 4 ArgonFew votes1512
- #2Claude Opus 5.51497
- #35Muse GlimmerFew votes1418
- ~#36Step 3.7 FlashEstimated1417
- #36MiMo-V2.51411
No published score — estimated from related benchmarks
LMArena Text (Elo)Estimated
~#36 of 57 (estimated)~1431 Elo
- #1Gemini 4 Argon1525
- #2Claude Opus 5.51507
- #35MiMo-V2.51434
- ~#36Step 3.7 FlashEstimated1431
- #36Mistral Medium 3.51426
No published score — estimated from related benchmarks
EQ-Bench Creative Writing v3 (Elo)Estimated
~#13 of 26 (estimated)~1828 Elo
- #1GPT-6 Astra2173
- #2Claude Fable 5.12162
- #12Qwen3.8 2.4T A95B1843
- ~#13Step 3.7 FlashEstimated1828
- #13Muse Glimmer1798
No published score — estimated from related benchmarks
Vision
MMMU-ProVerified
#36 of 8775.3%
- #1Claude Opus 5.587.7%
- #2GPT-6 Astra86.9%
- #35MiMo-V2.575.4%
- #36Step 3.7 Flash75.3%
- #37Qwen3.5 122B A10B75.0%
artificialanalysis.ai · 2026-09-24
CharXiv ReasoningEstimated
~#8 of 12 (estimated)~78.9%
- #1Gemini 3.8 Flash86.2%
- #2Kimi K384.8%
- #7Nemotron 3.5 Super VL NVFP479.9%
- ~#8Step 3.7 FlashEstimated78.9%
- #8Muse Glimmer78.8%
No published score — estimated from related benchmarks
OmniDocBench v1.5Estimated
~#5 of 8 (estimated)~85.1%
- #1MiniMax-M391.6%
- #2Kimi K391.1%
- #4Gemini 3.1 Pro85.3%
- ~#5Step 3.7 FlashEstimated85.1%
- #5Claude Haiku 4.579.6%
No published score — estimated from related benchmarks
GDP.pdf (AA)Estimated
~#26 of 48 (estimated)~14.0%
- #1Muse Wonder Full32.6%
- #2GPT-6 Astra31.0%
- #25Step 5 Preview14.8%
- ~#26Step 3.7 FlashEstimated14.0%
- #26Gemini 3.5 Flash-Lite13.6%
No published score — estimated from related benchmarks
ChartographyEstimated
~#16 of 25 (estimated)~21.5%
- #1Claude Fable 5.273.4%
- #2GPT-6 Astra71.0%
- #15Gemini 3.1 Pro26.1%
- ~#16Step 3.7 FlashEstimated21.5%
- #16Grok 4.620.5%
No published score — estimated from related benchmarks
Hard reasoning
GPQA DiamondVerified
#76 of 19280.9%
- #1GPT-6 Astra96.1%
- #2Gemini 3.8 Flash95.3%
- #74Nova 2.0 Lite81.1%
- #76Step 3.7 Flash80.9%
- #77Qwen3.5 9B80.6%
artificialanalysis.ai · 2026-09-24
Humanity's Last ExamVerified
#78 of 19821.4%
- #1Claude Fable 5.274.3%
- #2Claude Opus 5.561.4%
- #77Ring-2.6-1T21.6%
- #78Step 3.7 Flash21.4%
- #79Nemotron 3 Super 120B A12B20.8%
artificialanalysis.ai · 2026-09-24
AIME (latest)Estimated
~#19 of 34 (estimated)~85.3%
- #1GLM-5.3 NVFP499.5%
- #2Beam 1.0 501B A23B97.7%
- #18Mistral Medium 3.586.3%
- ~#19Step 3.7 FlashEstimated85.3%
- #19Mistral Small 484.0%
No published score — estimated from related benchmarks
AA Intelligence IndexEstimated
~#53 of 103 (estimated)~24.3
- #1Gemini 4 Carbon59.4
- #2Claude Opus 5.557.6
- #52Ling-3.0-flash-VL24.6
- ~#53Step 3.7 FlashEstimated24.3
- #53GPT-5.4 mini24.1
No published score — estimated from related benchmarks
Agents
OSWorld-VerifiedEstimated
~#11 of 15 (estimated)~66.9%
- #1Kimi K384.8%
- #2Qwen3.8 27B84.3%
- #10Nex-N2.5-Mini71.2%
- ~#11Step 3.7 FlashEstimated66.9%
- #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
BrowseCompEstimated
~#19 of 22 (estimated)~66.0%
- #1Atria Dawn Preview92.5%
- #2GPT-6 Astra91.5%
- #18Nemotron 3.5 Bolt 74B-A9B70.0%
- ~#19Step 3.7 FlashEstimated66.0%
- #19Mistral Medium 3.548.6%
No published score — estimated from related benchmarks
GDPval-AA (Elo)Verified
#50 of 91845 Elo
- #1Claude Opus 5.51866
- #2Muse Wonder Full1857
- #49LongCat 2.0856
- #50Step 3.7 Flash845
- #51Qwen3.5 122B A10B802
artificialanalysis.ai · 2026-09-24
AutomationBench-AAEstimated
~#45 of 85 (estimated)~17.0%
- #1Claude Fable 5.276.4%
- #2Muse Wonder Full71.3%
- #44Qwen3.7 Plus17.4%
- ~#45Step 3.7 FlashEstimated17.0%
- #45Ling-3.0-flash-VL15.7%
No published score — estimated from related benchmarks