DeepSeek V4.1 Flash NVFP4
- Input
- —
- Output
- —
- Speed
- ~160tok/sest.
- Context
- 1Mtokens
Rankings
Coding
Lab-reported#20 / 137rank by quality72.3±10.2quality / 1001/7 benchmarks measuredWriting & chat
Lab-reported#15 / 162rank by quality77.2±9.7quality / 1001/5 benchmarks measuredVision
Lab-reported#31 / 98rank by quality66.6±8.7quality / 1001/6 benchmarks measuredHard reasoning
Lab-reported#40 / 218rank by quality70.2±9.7quality / 1001/4 benchmarks measured
Self-hostingEstimated
How we estimate →What it would cost to run these open weights yourself on rented GPUs. No API sells this model, so this is its price on the chart.
- Hardware
- 4× B200 180GB
- NVFP4 weights
- Throughput
- ~17,975 tok/s
- many requests batched
- Price
- $0.15–1.32 /M tok
- busy → light use
- On your own machine
- Mac Studio M5 Ultra 512GB (NVFP4, ~93 tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- NVFP4 weights (552B params, 16B active per token) + 40% KV-cache headroom ≈ 435 GB
- 4× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
- ~17,975 output tok/s aggregate at batch 514 (bandwidth-bound); MoE compute scales with active params
- Blended 3:1 input:output; prefill ~265,781 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
~#2 of 21 (estimated)~86.1%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Estimated
~#6 of 24 (estimated)~70.5%
- #1Claude Fable 5.293.0%
- #2Claude Opus 5.589.9%
- #5Claude Opus 579.2%
- ~#6DeepSeek V4.1 Flash NVFP4Estimated70.5%
- #6Intern-S2-397B68.5%
No published score — estimated from related benchmarks
DeepSWE v1.1Estimated
~#24 of 46 (estimated)~63.9%
- #1Gemini 4 Argon78.8%
- #2DeepSeek V4.1 Flash 101777.0%
- #22Hy4 preview64.3%
- ~#24DeepSeek V4.1 Flash NVFP4Estimated63.9%
- #24GPT-6 Luna63.7%
No published score — estimated from related benchmarks
Terminal-Bench 4.0Estimated
~#24 of 102 (estimated)~30.5%
- #1Kimi K3.1 Pro67.2%
- #2Claude Sonnet 5.563.6%
- #22GLM-5.3 Flash32.8%
- ~#24DeepSeek V4.1 Flash NVFP4Estimated30.5%
- #24DeepSeek V4.1 Flash26.8%
No published score — estimated from related benchmarks
LiveCodeBenchEstimated
~#3 of 28 (estimated)~90.7%
- #1Qwen3.8-Omni-Flash92.6%
- #2Qwen3.8 Flash-Next91.9%
- ~#3DeepSeek V4.1 Flash NVFP4Estimated90.7%
- #3Qwen3.8 27B90.3%
No published score — estimated from related benchmarks
SciCodeLab-reported
#24 of 11855.8%
- #1Claude Fable 5.267.7%
- #2Claude Opus 5.566.9%
- #23DeepSeek V4.1 Flash 101755.9%
- #24DeepSeek V4.1 Flash NVFP455.8%
- #25GPT-5.555.8%
huggingface.co · 2026-09-29
AA Coding Agent IndexEstimated
~#11 of 26 (estimated)~49.6
- #1Claude Opus 5.561.7
- #2Claude Opus 559.7
- #10Muse Wonder52.0
- ~#11DeepSeek V4.1 Flash NVFP4Estimated49.6
- #11GLM-5.348.9
No published score — estimated from related benchmarks
Writing & chat
IFBenchLab-reported
#17 of 13977.3%
- #1MiniMax-M382.9%
- #2Qwen3.8 2.4T A95B82.8%
- #16Qwen3.7 Plus78.0%
- #17DeepSeek V4.1 Flash NVFP477.3%
- #18Gemini 3.1 Pro77.1%
huggingface.co · 2026-09-29
LMArena Text, Non-English (Elo)Estimated
~#17 of 57 (estimated)~1459 Elo
- #1Gemini 4 ArgonFew votes1512
- #2Claude Opus 5.51497
- #16DeepSeek V4.1 Flash1459
- ~#17DeepSeek V4.1 Flash NVFP4Estimated1459
- #17GPT-5.6 Terra1456
No published score — estimated from related benchmarks
LMArena Text (Elo)Estimated
~#17 of 57 (estimated)~1470 Elo
- #1Gemini 4 Argon1525
- #2Claude Opus 5.51507
- #16DeepSeek V4.1 Flash1474
- ~#17DeepSeek V4.1 Flash NVFP4Estimated1470
- #17GPT-5.6 Terra1466
No published score — estimated from related benchmarks
EQ-Bench Creative Writing v3 (Elo)Estimated
~#9 of 26 (estimated)~1987 Elo
- #1GPT-6 Astra2173
- #2Claude Fable 5.12162
- #8Grok 4.72007
- ~#9DeepSeek V4.1 Flash NVFP4Estimated1987
- #9Muse Spark 1.31906
No published score — estimated from related benchmarks
Vision
MMMU-ProLab-reported
#42 of 8773.7%
- #1Claude Opus 5.587.7%
- #2GPT-6 Astra86.9%
- #41Inkling Small74.0%
- #42DeepSeek V4.1 Flash NVFP473.7%
- #43Inkling73.5%
huggingface.co · 2026-09-29
CharXiv ReasoningEstimated
~#6 of 12 (estimated)~80.8%
- #1Gemini 3.8 Flash86.2%
- #2Kimi K384.8%
- #5MiniMax-M3.1-Flash82.5%
- ~#6DeepSeek V4.1 Flash NVFP4Estimated80.8%
- #6Nemotron 3.5 Super VL80.8%
No published score — estimated from related benchmarks
OmniDocBench v1.5Estimated
~#4 of 8 (estimated)~86.8%
- #1MiniMax-M391.6%
- #2Kimi K391.1%
- #2Qwen3.8 27B91.1%
- ~#4DeepSeek V4.1 Flash NVFP4Estimated86.8%
- #4Gemini 3.1 Pro85.3%
No published score — estimated from related benchmarks
GDP.pdf (AA)Estimated
~#20 of 48 (estimated)~18.3%
- #1Muse Wonder Full32.6%
- #2GPT-6 Astra31.0%
- #19MiMo-V2.6-Pro19.2%
- ~#20DeepSeek V4.1 Flash NVFP4Estimated18.3%
- #20Gemini 3.1 Pro17.8%
No published score — estimated from related benchmarks
ChartographyEstimated
~#10 of 25 (estimated)~29.5%
- #1Claude Fable 5.273.4%
- #2GPT-6 Astra71.0%
- #9GPT-5.531.0%
- ~#10DeepSeek V4.1 Flash NVFP4Estimated29.5%
- #10GPT-6 Luna29.1%
No published score — estimated from related benchmarks
Hard reasoning
GPQA DiamondLab-reported
#24 of 19291.3%
- #1GPT-6 Astra96.1%
- #2Gemini 3.8 Flash95.3%
- #22Qwen3.8 Flash-Next NVFP491.5%
- #24DeepSeek V4.1 Flash NVFP491.3%
- #25GLM-5.3 Flash91.2%
huggingface.co · 2026-09-29
Humanity's Last ExamEstimated
~#53 of 199 (estimated)~32.0%
- #1Claude Fable 5.274.3%
- #2Claude Opus 5.561.4%
- #52Inkling Small33.3%
- ~#53DeepSeek V4.1 Flash NVFP4Estimated32.0%
- #53K2 Horizon 375B A23B32.0%
No published score — estimated from related benchmarks
AIME (latest)Estimated
~#2 of 34 (estimated)~99.0%
No published score — estimated from related benchmarks
AA Intelligence IndexEstimated
~#36 of 103 (estimated)~36.4
- #1Gemini 4 Carbon59.4
- #2Claude Opus 5.557.6
- #35GPT-6 Luna38.1
- ~#36DeepSeek V4.1 Flash NVFP4Estimated36.4
- #36DeepSeek V4 Pro (0813)36.0
No published score — estimated from related benchmarks