priorsBuilder

DeepSeek · released 2026-08-03

DeepSeek V4 Flash

Measured on 37 benchmarks across 4 reasoning efforts.

General frontier

41.1

#55 of 136 models · best at max

includes estimates: 31% of the mix is measured for this model

Reasoning efforts in your mix

max41.1#134 of 313
none33.9#194 of 313
max33.0#201 of 313
xhigh31.7#209 of 313

Your mix’s benchmarks

5 of 8 measured

reasoning

Humanity's Last Exam

587 results
  1. #122max of 58734.5%

Leader: Opus 5.5 · 61.4%

knowledge

AA-Omniscience

526 results
  1. #226max of 526-17.6

Leader: Opus 5.5 · 46.4

reasoning

BullshitBench v2

196 results
  1. #152none of 19618.0%
  2. #164xhigh of 19614.0%

Leader: Opus 4.8 · 95.0%

coding

DeepSWE v1.1

70 results
  1. #46max of 7053.3%

Leader: GPT-6 Astra · 74.1%

reasoning

SimpleBench

97 results
  1. #34max of 9761.1%

Leader: Opus 5.5 · 88.4%

All 37 benchmarks

show 32 more

agentic

AA-Briefcase v1.1

224 results
  1. #52max of 2241425

Leader: Sonnet 5.5 · 1823

agentic

AutomationBench (Zapier)

115 results
  1. #42max of 11518.1%

Leader: Gemini 4 · 51.3%

agentic

AutomationBench-AA

219 results
  1. #82max of 21947.5%

Leader: Gemini 4 · 77.5%

agentic

EnterpriseOps-Gym-AA

50 results
  1. #11max of 5047.1%

Leader: Fable 5 · 51.1%

agentic

τ³-Banking

206 results
  1. #28max of 20641.0%

Leader: Qwen 3.8 Max · 51.3%

coding

Arena WebDev

122 results
  1. #32high of 1221582
  2. #75max of 1221431

Leader: Opus 5.5 · 1813

coding

FrontierSWE v2

21 results
  1. #19max of 2114.8%

Leader: GPT-6 Astra · 65.5%

coding

SciCode

226 results
  1. #110max of 22649.7%

Leader: Opus 5.5 · 66.9%

coding

Terminal-Bench v2.1-AA

238 results
  1. #65max of 23874.2%

Leader: Fable 5.1 · 91.4%

coding

Terminal-Bench v4-AA

216 results
  1. #81max of 21612.1%

Leader: Sonnet 5.5 · 63.6%

coding

WeirdML v2

127 results
  1. #73max of 12745.6%

Leader: GPT-6 Astra Pro · 93.6%

composite

AA Intelligence Index v4.3.2

610 results
  1. #85max of 61034.8

Leader: Opus 5.5 · 57.6

composite

Arena Text

220 results
  1. #82max of 2201439

Leader: Gemini 4 · 1525

composite

Arena Text — Multi-Turn

220 results
  1. #70max of 2201453

Leader: Gemini 4 · 1553

cost

AA Intelligence Index Cost per Task

177 results
  1. #70max of 177$0.31

Leader: GPT-6 Luna · $0.00

cost

Cost to Run AA Intelligence Index

177 results
  1. #63max of 177$445

Leader: GPT-6 Luna · $10.63

creative

VoxelBench (text)

55 results
  1. #38max of 551412

Leader: GPT-6 Astra · 2732

economics

GDP.pdf

210 results
  1. #112max of 21012.0%

Leader: Muse Wonder Full (Max) · 32.6%

economics

GDPval-AA v2.1

288 results
  1. #42max of 2881547

Leader: Opus 5.5 · 1866

knowledge

AA-Omniscience Accuracy

526 results
  1. #130max of 52638.6%

Leader: Fable 5.1 · 67.2%

knowledge

AA-Omniscience Non-hallucination

526 results
  1. #436max of 5268.5%

Leader: MiniCPM5-1B (Non-reasoning) · 99.1%

knowledge

GPQA Diamond

539 results
  1. #48max of 53991.3%

Leader: GPT-6 Astra · 96.3%

knowledge

MMMU-Pro

265 results
  1. #103max of 26574.8%

Leader: Opus 5.5 · 87.7%

long-context

AA-LCR v1.1

523 results
  1. #57max of 52381.3%

Leader: Kimi K3 · 88.7%

reasoning

CritPt

529 results
  1. #109max of 52910.9%

Leader: Muse Wonder Full (Max) · 32.9%

reasoning

Roboflow Visual Reasoning

102 results
  1. #81high of 10249.0%
  2. #99low of 10231.8%

Leader: GPT-6 Astra · 91.2%

reliability

Tool-call error rate

171 results
  1. #102max of 1711.9%

Leader: GPT-5.4 Pro · 0.0%

reliability

Uptime (7d)

181 results
  1. #136max of 18198.4%

Leader: GPT-5 Pro · 100.0%

speed

Latency (Time to First Token)

351 results
  1. #75max of 3511.0s

Leader: Gemini 2.5 Flash-Lite · 0.3s

speed

Output Speed (tokens/s)

351 results
  1. #52max of 351188.9

Leader: Celeris-1 · 1522.5

speed

Provider latency

177 results
  1. #109max of 1771.6s

Leader: Qwen 2.5 7B · 0.2s

speed

Provider throughput

176 results
  1. #35max of 17697.0

Leader: Muse Spark 1.1 · 204.3