Skip to content
Model Pareto

Step 3.7 Flash

StepFunOpen weightsartificialanalysis.ai ↗
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

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

~#13 of 21 (estimated)~73.4%
  1. #1Claude Opus 596.0%
  2. #2Claude Sonnet 585.2%
  3. #12Flint74.2%
  4. ~#13Step 3.7 FlashEstimated73.4%
  5. #13Claude Haiku 4.573.3%
No published score — estimated from related benchmarks
LiveCodeBenchEstimated
~#15 of 28 (estimated)~69.9%
  1. #1Qwen3.8-Omni-Flash92.6%
  2. #2Qwen3.8 Flash-Next91.9%
  3. #14Maple-Preview75.1%
  4. ~#15Step 3.7 FlashEstimated69.9%
  5. #15Mistral Small 464.0%
No published score — estimated from related benchmarks
~#23 of 26 (estimated)~34.3
  1. #1Claude Opus 5.561.7
  2. #2Claude Opus 559.7
  3. #22Claude Haiku 5.537.3
  4. ~#23Step 3.7 FlashEstimated34.3
  5. #23Gemini 3.8 Flash33.3
No published score — estimated from related benchmarks

Writing & chat

Vision

MMMUEstimated
~#2 of 3 (estimated)~71.3%
  1. #1Claude Haiku 4.573.2%
  2. ~#2Step 3.7 FlashEstimated71.3%
  3. #2Llama 4 Maverick69.4%
No published score — estimated from related benchmarks
~#8 of 12 (estimated)~78.9%
  1. #1Gemini 3.8 Flash86.2%
  2. #2Kimi K384.8%
  3. #7Nemotron 3.5 Super VL NVFP479.9%
  4. ~#8Step 3.7 FlashEstimated78.9%
  5. #8Muse Glimmer78.8%
No published score — estimated from related benchmarks
~#5 of 8 (estimated)~85.1%
  1. #1MiniMax-M391.6%
  2. #2Kimi K391.1%
  3. #4Gemini 3.1 Pro85.3%
  4. ~#5Step 3.7 FlashEstimated85.1%
  5. #5Claude Haiku 4.579.6%
No published score — estimated from related benchmarks
GDP.pdf (AA)Estimated
~#26 of 48 (estimated)~14.0%
  1. #1Muse Wonder Full32.6%
  2. #2GPT-6 Astra31.0%
  3. #25Step 5 Preview14.8%
  4. ~#26Step 3.7 FlashEstimated14.0%
  5. #26Gemini 3.5 Flash-Lite13.6%
No published score — estimated from related benchmarks
ChartographyEstimated
~#16 of 25 (estimated)~21.5%
  1. #1Claude Fable 5.273.4%
  2. #2GPT-6 Astra71.0%
  3. #15Gemini 3.1 Pro26.1%
  4. ~#16Step 3.7 FlashEstimated21.5%
  5. #16Grok 4.620.5%
No published score — estimated from related benchmarks

Hard reasoning

Agents