NEWHow AI models draw the world map · Gecko Tests
Beta
Home/Models/DeepSeek V4 Flash 0731
DeepSeek logo

DeepSeek V4 Flash 0731

by DeepSeek · Released Jul 2026

1M Context
59.9
avg score
Rank #93
Compare
Better than 71% of all models
Context
1.0M tokens (~524 books)
Input $/1M
$0.07
Output $/1M
$0.14
Type
text
License
Proprietary
Benchmarks
15 tested
Data as of
About

DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows....

Tested on 15 benchmarks · BenchGecko score 59.9. Top scores: OTIS Mock AIME 2024-2025 (94.4%), ARC-AGI (89.0%), GPQA diamond (88.0%).

Capabilities
coding
40.9
#136 globally
reasoning
67.9
#53 globally
math
58.8
#96 globally
knowledge
50.0
#133 globally
general
54.3
#25 globally
Benchmark Scores
Compare All
Tested on 15 benchmarks · Ranked across 5 categories
Score Distribution (all 319 models)
0255075100
▲ You are here
codingCompare coding →
WeirdML

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

63.0·
Frontiercode
18.8·
reasoningCompare reasoning →
ARC-AGI

Abstraction and Reasoning Corpus. Tests fluid intelligence through novel visual pattern recognition puzzles. Core measure of general intelligence.

89.0·
ARC-AGI-2

ARC-AGI 2, harder sequel to ARC. More complex abstract reasoning patterns that test generalization ability beyond training data.

61.4·
SimpleBench

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

53.3·
mathCompare math →
OTIS Mock AIME 2024-2025

Mock AIME (American Invitational Mathematics Exam) problems from OTIS. Tests mathematical competition performance.

94.4·
FrontierMath-Tiers-1-3-v2-Private
57.5·
FrontierMath-Tier-4-v2-Private
24.4·
Quick compare:Google DeepMind logovs Gemini 3.6 FlashAlibaba Qwen logovs Qwen3.8 Max (0902)vs Stable Beluga 2
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Model Family · DeepSeek DeepSeek V4
Apr 2026
68.6
$0.09/M in1.0Mctx16 benchmarks
Jul 2026
59.9-8.7
$0.02/M in(-0.07)1.0Mctx15 benchmarks
Jul 2026
$0.14/M in(+0.12)1.0Mctx
Jul 2026
$0.00/M in(-0.14)1.0Mctx
Aug 2026
$0.22/M in(+0.22)1.0Mctx
Aug 2026
$0.11/M in(-0.11)1.0Mctx
Apr 2026
60.6+0.7
$0.96/M in(+0.85)1.0Mctx34 benchmarks
Aug 2026
66.1+5.5
$0.66/M in(-0.30)1.0Mctx21 benchmarks
Aug 2026
$0.66/M in1.0Mctx
See the full DeepSeek V4 family →
Recently Happened
DeepSeek V4 Flash 0731 pricing increased 132%
Oct 15, 2026
DeepSeek V4 Flash 0731 pricing increased 198%
Oct 4, 2026
DeepSeek V4 Flash 0731 pricing dropped 14%
Sep 29, 2026
DeepSeek V4 Flash 0731 pricing dropped 30%
Sep 26, 2026
DeepSeek V4 Flash 0731 pricing increased 300%
Sep 22, 2026
DeepSeek V4 Flash 0731 pricing increased 20%
Sep 16, 2026
Similar Models
Google DeepMind logo
Gemini 3.6 Flash
Google DeepMind
60.0$0.75/1M
Alibaba Qwen logo
Qwen3.8 Max (0902)
Alibaba Qwen
59.7$2.00/1M
Stable Beluga 2
Unknown
60.1n/a
Links
Info
DeepSeekPricing explorerDevelopers · API
Research
Technical Report
Documentation
API DocsPlayground
Community
@DeepSeek
BenchGecko API
deepseek-v4-flash-0731
Specifications
  • Typetext
  • Context1.0M tokens (~524 books)
  • ReleasedJul 2026
  • LicenseProprietary
  • StatusActive
  • Cost / Message~$0.000
Available On
DeepSeek logoDeepSeek$0.07
Categories
codingreasoningmathknowledgegeneral
Learn More
context-windowtransformertokens
Share & Export
Tweet
Related Models
Google DeepMind logoGemini 3.6 Flash
60.0
Alibaba Qwen logoQwen3.8 Max (0902)
59.7
Stable Beluga 2
60.1
Alibaba Qwen logoQwen-14B
60.1
DeepSeek V4 Flash 0731 is a proprietary text AI model by DeepSeek, released in July 2026. It has an average benchmark score of 59.9. Context window: 1M tokens.

Related Models

Gemini 3.6 Flash · Google DeepMindQwen3.8 Max (0902) · Alibaba QwenStable Beluga 2 · UnknownQwen-14B · Alibaba QwenGPT-5 · OpenAI

Benchmarks

OTIS Mock AIME 2024-2025ARC-AGIGPQA diamondDtbenchWeirdML

Related Pages

DeepSeek · ProviderDeepSeek · EconomyAll ModelsCompare ModelsPricingDevelopers · APIContext Window · Glossary

Key facts · as of 2026-10-17

  • DeepSeek V4 Flash 0731 by DeepSeek. BenchGecko score 59.9, rank 93 of 319 scored models (normalized average of public benchmark scores).
  • List price $0.0700 input · $0.14 output per 1M tokens (as of 2026-10-17).
  • Sold by 27 providers (as of 2026-10-17): OpenInference (fp4) $0.0171 in / $0.97 out · Relace (fp4) $0.0171 in / $0.97 out · Reka $0.0210 in / $0.53 out · Baidu (fp8) $0.0598 in / $0.18 out · DeepInfra (fp8) $0.0600 in / $0.18 out · and 22 more. Every provider
  • Gecko Tests · Gecko Score 22 (rank 27): Who Are You E (Sometimes says it is Google or Anthropic) · Knowledge Horizon D (Knows news up to Jun 2025) · Tokenizer Tax D (68% more tokens outside English).

How to cite · data as of 2026-10-17

DeepSeek V4 Flash 0731 · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-17. https://benchgecko.ai/model/deepseek-v4-flash-0731

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