DataLearner logo

DeepSeek-V4-Flash-Vision-ExpvsDeepSeek-V4-Flash

Across 5 shared benchmarks, DeepSeek-V4-Flash-Vision-Exp leads overall: DeepSeek-V4-Flash-Vision-Exp wins 5, DeepSeek-V4-Flash wins 0, with 0 ties and an average score difference of +2.46.

DeepSeek-AI
DeepSeek-V4-Flash-Vision-Exp

DeepSeek-AI · 2026-08-21 · Multimodal model

DeepSeek-AI
DeepSeek-V4-Flash

DeepSeek-AI · 2026-04-24 · Reasoning model

DeepSeek-V4-Flash-Vision-Exp5 wins(100%)(0%)0 winsDeepSeek-V4-Flash

Benchmark scores

Grouped by capability, sorted by largest gap within each. 5 shared benchmarks.

AI Agent - Tool Usage

DeepSeek-V4-Flash-Vision-Exp 2/2
BenchmarkDeepSeek-V4-Flash-Vision-ExpDeepSeek-V4-FlashDiff
Terminal-Bench 2.183.9011 / 44Max (With Tools)82.7016 / 44Max (With Tools)+1.20
AutomationBench25.707 / 8Max (With Tools)25.108 / 8Max (With Tools)+0.60

Coding and Software Engineer

DeepSeek-V4-Flash-Vision-Exp 2/2
BenchmarkDeepSeek-V4-Flash-Vision-ExpDeepSeek-V4-FlashDiff
DeepSWE59.3012 / 27Max (With Tools)54.4015 / 27Max (With Tools)+4.90
NL2Repo-Bench57.703 / 8Max (With Tools)54.205 / 8Max (With Tools)+3.50

Agent Level Benchmark

DeepSeek-V4-Flash-Vision-Exp 1/1
BenchmarkDeepSeek-V4-Flash-Vision-ExpDeepSeek-V4-FlashDiff
Agents' Last Exam27.306 / 11Max (With Tools)25.2010 / 11Max (With Tools)+2.10

Specs

FieldDeepSeek-V4-Flash-Vision-ExpDeepSeek-V4-Flash
PublisherDeepSeek-AIDeepSeek-AI
Release date2026-08-212026-04-24
Model typeMultimodal modelReasoning model
ArchitectureDenseMoE
ParametersNot available284B
Context length1M1M
Max output384K384K

API pricing

Prices use DataLearner records when available; missing fields are not inferred.

ItemDeepSeek-V4-Flash-Vision-ExpDeepSeek-V4-Flash
Text input$0.22 / 1M tokens$0.14 / 1M tokens
Text output$0.66 / 1M tokens$0.28 / 1M tokens
Cache read$0.007 / 1M tokens$0.0028 / 1M tokens

Summary

  • DeepSeek-V4-Flash-Vision-Expleads in:AI Agent - Tool Usage (2/2), Coding and Software Engineer (2/2), Agent Level Benchmark (1/1)

On average across the 5 shared benchmarks, DeepSeek-V4-Flash-Vision-Exp scores 2.46 higher.

Largest single-benchmark gap: DeepSWE — DeepSeek-V4-Flash-Vision-Exp 59.30 vs DeepSeek-V4-Flash 54.40 (+4.90).

Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.