Qwen3.5 397B A17B
- Input
- $0.60/M tok
- Output
- $3.60/M tok
- Blended
- $1.35/M tok · 3:1
- Speed
- 86tok/s
- Context
- 262Ktokens
- Cost per task
- ~$1.5/taskest.
Long-tail import from Artificial Analysis (2026-09-24); AA variant: Qwen3.5 397B A17B (Reasoning). Price = AA median host price.
Rankings
Coding
Verified#82 / 137rank by quality39.1±8.1quality / 1002/7 benchmarks measuredWriting & chat
Verified#38 / 162rank by quality66.4±4.1quality / 1003/5 benchmarks measuredVision
Verified#44 / 98rank by quality58.1±6.3quality / 1002/6 benchmarks measuredHard reasoning
Verified#77 / 218rank by quality54.2±3.7quality / 1003/4 benchmarks measuredAgents
Verified#82 / 187rank by quality37.0±4.3quality / 1004/7 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
- 4× B200 180GB
- FP8 weights
- Throughput
- ~12,969 tok/s
- many requests batched
- Price
- $0.20–1.76 /M tok
- busy → light use
- On your own machine
- Mac Studio M5 Ultra 512GB (int4, ~99 tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- FP8 weights (397B params, 17B active per token) + 25% KV-cache headroom ≈ 496 GB
- 4× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
- ~12,969 output tok/s aggregate at batch 371 (bandwidth-bound); MoE compute scales with active params
- Blended 3:1 input:output; prefill ~250,147 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)~73.7%
- #1Claude Opus 596.0%
- #2Claude Sonnet 585.2%
- #12Flint74.2%
- ~#13Qwen3.5 397B A17BEstimated73.7%
- #13Claude Haiku 4.573.3%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Estimated
~#20 of 24 (estimated)~53.4%
- #1Claude Fable 5.293.0%
- #2Claude Opus 5.589.9%
- #18Gemini 3.5 Flash-Lite54.2%
- ~#20Qwen3.5 397B A17BEstimated53.4%
- #20Nemotron 3.5 Super VL51.8%
No published score — estimated from related benchmarks
DeepSWE v1.1Estimated
~#38 of 46 (estimated)~48.7%
- #1Gemini 4 Argon78.8%
- #2DeepSeek V4.1 Flash 101777.0%
- #37Qwen3.8 Max (0902)51.0%
- ~#38Qwen3.5 397B A17BEstimated48.7%
- #38Ripple-65B Preview44.3%
No published score — estimated from related benchmarks
Terminal-Bench 4.0Verified
#63 of 1010.0%
- #1Kimi K3.1 Pro67.2%
- #2Claude Sonnet 5.563.6%
- #63Qwen3.5 122B A10B0.0%
- #63Qwen3.5 397B A17B0.0%
- #63Qwen3.6 35B A3B0.0%
artificialanalysis.ai · 2026-09-24
LiveCodeBenchEstimated
~#15 of 28 (estimated)~68.3%
- #1Qwen3.8-Omni-Flash92.6%
- #2Qwen3.8 Flash-Next91.9%
- #14Maple-Preview75.1%
- ~#15Qwen3.5 397B A17BEstimated68.3%
- #15Mistral Small 464.0%
No published score — estimated from related benchmarks
SciCodeVerified
#64 of 11844.8%
- #1Claude Fable 5.267.7%
- #2Claude Opus 5.566.9%
- #63Muse Glimmer44.9%
- #64Qwen3.5 397B A17B44.8%
- #65Ling-3.0-flash-VL44.2%
artificialanalysis.ai · 2026-09-24
AA Coding Agent IndexEstimated
~#24 of 26 (estimated)~32.6
- #1Claude Opus 5.561.7
- #2Claude Opus 559.7
- #23Gemini 3.8 Flash33.3
- ~#24Qwen3.5 397B A17BEstimated32.6
- #24GPT-6 Luna31.1
No published score — estimated from related benchmarks
Writing & chat
IFBenchVerified
#15 of 13978.8%
- #1MiniMax-M382.9%
- #2Qwen3.8 2.4T A95B82.8%
- #14Qwen3.8 27B NVFP478.9%
- #15Qwen3.5 397B A17B78.8%
- #16Qwen3.7 Plus78.0%
artificialanalysis.ai · 2026-09-24
LMArena Text, Non-English (Elo)Verified
#28 of 561434 Elo
- #1Gemini 4 ArgonFew votes1512
- #2Claude Opus 5.51497
- #27GPT-6 Luna1436
- #28Qwen3.5 397B A17B1434
- #29Step 5 Preview1433
lmarena.ai · 2026-09-13
EQ-Bench Creative Writing v3 (Elo)Estimated
~#13 of 26 (estimated)~1809 Elo
- #1GPT-6 Astra2173
- #2Claude Fable 5.12162
- #12Qwen3.8 2.4T A95B1843
- ~#13Qwen3.5 397B A17BEstimated1809
- #13Muse Glimmer1798
No published score — estimated from related benchmarks
Vision
MMMU-ProVerified
#26 of 8777.3%
- #1Claude Opus 5.587.7%
- #2GPT-6 Astra86.9%
- #26Claude Sonnet 577.3%
- #26Qwen3.5 397B A17B77.3%
- #28DeepSeek V4.1 Flash77.0%
artificialanalysis.ai · 2026-09-24
CharXiv ReasoningEstimated
~#9 of 12 (estimated)~78.7%
- #1Gemini 3.8 Flash86.2%
- #2Kimi K384.8%
- #8Muse Glimmer78.8%
- ~#9Qwen3.5 397B A17BEstimated78.7%
- #9Inkling78.1%
No published score — estimated from related benchmarks
OmniDocBench v1.5Estimated
~#5 of 8 (estimated)~84.7%
- #1MiniMax-M391.6%
- #2Kimi K391.1%
- #4Gemini 3.1 Pro85.3%
- ~#5Qwen3.5 397B A17BEstimated84.7%
- #5Claude Haiku 4.579.6%
No published score — estimated from related benchmarks
GDP.pdf (AA)Verified
#31 of 4712.2%
- #1Muse Wonder Full32.6%
- #2GPT-6 Astra31.0%
- #29Inkling12.8%
- #31Qwen3.5 397B A17B12.2%
- #31Qwen3.7 Plus12.2%
artificialanalysis.ai · 2026-09-24
ChartographyEstimated
~#17 of 25 (estimated)~19.0%
- #1Claude Fable 5.273.4%
- #2GPT-6 Astra71.0%
- #16Grok 4.620.5%
- ~#17Qwen3.5 397B A17BEstimated19.0%
- #17Muse Glimmer17.2%
No published score — estimated from related benchmarks
Hard reasoning
GPQA DiamondVerified
#38 of 19289.3%
- #1GPT-6 Astra96.1%
- #2Gemini 3.8 Flash95.3%
- #36Inkling Small89.5%
- #38Qwen3.5 397B A17B89.3%
- #39Nex-N2-Pro89.2%
artificialanalysis.ai · 2026-09-24
Humanity's Last ExamVerified
#59 of 19829.0%
- #1Claude Fable 5.274.3%
- #2Claude Opus 5.561.4%
- #58Solar Pro 429.2%
- #59Qwen3.5 397B A17B29.0%
- #60Solar Open2 250B28.5%
artificialanalysis.ai · 2026-09-24
AIME (latest)Estimated
~#10 of 34 (estimated)~92.2%
- #1GLM-5.3 NVFP499.5%
- #2Beam 1.0 501B A23B97.7%
- #9Qwen3.8 27B MXFP492.5%
- ~#10Qwen3.5 397B A17BEstimated92.2%
- #10GLM-5.3-MXFP491.3%
No published score — estimated from related benchmarks
Agents
τ²-bench (Telecom)Verified
#3 of 11995.6%
- #1Step 3.7 Flash98.5%
- #2DeepSeek-V4-Pro-0813-nvfp4-DSpark98.3%
- #3Gemini 3.1 Pro95.6%
- #3Qwen3.5 397B A17B95.6%
- #5Qwen3.6 35B A3B95.3%
artificialanalysis.ai · 2026-09-24
OSWorld-VerifiedEstimated
~#12 of 15 (estimated)~65.3%
- #1Kimi K384.8%
- #2Qwen3.8 27B84.3%
- #11Muse Glimmer65.9%
- ~#12Qwen3.5 397B A17BEstimated65.3%
- #12Nemotron 3.5 Super VL NVFP452.3%
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)~64.2%
- #1Atria Dawn Preview92.5%
- #2GPT-6 Astra91.5%
- #18Nemotron 3.5 Bolt 74B-A9B70.0%
- ~#19Qwen3.5 397B A17BEstimated64.2%
- #19Mistral Medium 3.548.6%
No published score — estimated from related benchmarks
GDPval-AA (Elo)Verified
#53 of 91780 Elo
- #1Claude Opus 5.51866
- #2Muse Wonder Full1857
- #52Muse Glimmer786
- #53Qwen3.5 397B A17B780
- #54Gemini 3.1 Pro776
artificialanalysis.ai · 2026-09-24