DeepSeek logo

DeepSeek V4 Flash

DeepSeek · Released Apr 24, 2026

Open SourcebudgetcodingCompare this model

$0.09

Input price / MTok

#2 of 27

$0.18

Output price / MTok

#1 of 27

1M

Context window

#16 of 28

113

Tokens / sec

#10 of 21

88.1

GPQA Diamond

#9 of 19

79

SWE-bench Verified

#5 of 12

91.6

LiveCodeBench

#3 of 8

Summary

DeepSeek V4 Flash is a open-source language model from DeepSeek, released on Apr 24, 2026.

At $0.09 input and $0.18 output per million tokens, it ranks #2 of 27 for input price, placing it among the most affordable options.

Its context window of 1M tokens ranks #16 of 28 for context depth in this dataset.

It targets budget, coding workloads.

Throughput is measured at 113 tokens per second (ranked #10 of 21 for throughput).

284B total params / 13B active; the deepseek-reasoner API alias now routes here.

Benchmarks

BenchmarkScoreMax%RankSource
GPQA Diamond88.110088.1%#9/19DeepSeek (self-reported in technical rep...
SWE-bench Verified7910079.0%#5/12DeepSeek (self-reported in technical rep...
LiveCodeBench91.610091.6%#3/8DeepSeek (self-reported in technical rep...

Similar Models

Compare all 6 models

Models with the most shared benchmark coverage and closest scores to DeepSeek V4 Flash.

DeepSeek V4 Flash
DeepSeekDeepSeek V4 Pro
GoogleGemini 2.5 Pro
GoogleGemini 2.5 Flash
AnthropicClaude Sonnet 4.6
TencentTencent Hy3
GPQA Diamond#9 of 19
Graduate-level science reasoning
SWE-bench Verified#5 of 12
Real-world GitHub issue resolution
LiveCodeBench#3 of 8
Live competitive programming

Compare DeepSeek V4 Flash side by side

Pre-filled with 5 similar models

Open comparison

Speed source: https://artificialanalysis.ai/models/deepseek-v4-flash