Home/Models/DeepSeek V4 Flash
DeepSeek logo

DeepSeek V4 Flash

by DeepSeek · Released Apr 2026

Open Source1M Context
68.6
avg score
Rank #43
Compare
Better than 86% of all models
Context
1.0M tokens (~524 books)
Input $/1M
$0.03
Output $/1M
$1.28
Type
text
License
Open Source
Benchmarks
16 tested
Data as of
About

DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model from DeepSeek with 284B total parameters and 13B activated parameters, supporting a 1M-token context window. It is designed for fast inference and...

Tested on 16 benchmarks · BenchGecko score 68.6. Top scores: Chatbot Arena Elo — Overall (1436.0%), LiveBench — Mathematics (79.7%), Dtbench (77.3%).

Capabilities
coding
55.0
#71 globally
reasoning
58.1
#61 globally
math
79.7
#27 globally
knowledge
67.3
#27 globally
speed
42.5
#52 globally
language
66.6
#92 globally
general
59.8
#14 globally
Benchmark Scores
Compare All
Tested on 16 benchmarks · Ranked across 8 categories
Score Distribution (all 312 models)
0255075100
▲ You are here
LiveBench — Coding

Regularly refreshed coding problems that avoid data contamination. New problems added monthly to prevent memorization.

69.2·
LiveBench — Agentic Coding

LiveBench coding tasks that require multi-step reasoning and tool use. Tests planning and execution of complex coding workflows.

50.0·
WeirdML

Unusual and adversarial machine learning challenges. Tests robustness of reasoning about edge cases in ML systems.

45.6·
LiveBench — Reasoning

Regularly refreshed reasoning problems testing logical deduction, spatial reasoning, and analytical thinking.

70.6·
LiveBench — Data Analysis

Fresh data analysis tasks testing ability to interpret tables, charts, and statistical data.

68.0·
SimpleBench

Deceptively simple questions that humans find easy but AI models often get wrong. Tests common sense and reasoning gaps.

35.6·
LiveBench — Mathematics

Regularly updated math problems that test numerical reasoning, algebra, calculus, and combinatorics.

79.7·
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Recently Happened
DeepSeek V4 Flash pricing dropped 70%
Oct 1, 2026
DeepSeek V4 Flash pricing increased 198%
Sep 28, 2026
DeepSeek V4 Flash pricing increased 142%
Sep 21, 2026
DeepSeek V4 Flash pricing increased 36%
Sep 14, 2026
DeepSeek V4 Flash pricing increased 8%
Sep 7, 2026
DeepSeek V4 Flash pricing increased 9%
Sep 2, 2026
Links
Documentation
Community
BenchGecko API
deepseek-v4-flash
Specifications
  • Typetext
  • Context1.0M tokens (~524 books)
  • ReleasedApr 2026
  • LicenseOpen Source
  • StatusActive
  • Cost / Message~$0.001
Available On
DeepSeek logoDeepSeek$0.03
Share & Export
Tweet
DeepSeek V4 Flash is an open-source text AI model by DeepSeek, released in April 2026. It has an average benchmark score of 68.6. Context window: 1M tokens.

Key facts · as of 2026-10-05

  • DeepSeek V4 Flash by DeepSeek. BenchGecko score 68.6, rank 43 of 312 scored models (normalized average of public benchmark scores).
  • List price $0.0250 input · $1.28 output per 1M tokens (as of 2026-10-05).
  • Sold by 16 providers (as of 2026-10-05): Relace (fp4) $0.0300 in / $1.28 out · OpenInference (fp4) $0.0390 in / $1.66 out · StreamLake (fp8) $0.0420 in / $0.0840 out · DeepInfra (fp8) $0.0900 in / $0.18 out · GMICloud (fp8) $0.0910 in / $0.18 out · and 11 more. Every provider
  • Gecko Tests: Who Are You B (Knows who made it) · Tokenizer Tax A (7% more tokens outside English).

How to cite · data as of 2026-10-05

DeepSeek V4 Flash · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-05. https://benchgecko.ai/model/deepseek-v4-flash

Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP