DeepSeek V3.1 vs DeepSeek V4-Flash

Best cell per row highlighted. Null means undisclosed — never counted as zero.

DeepSeek
DeepSeek V3.1GA
7.7
AI Panel
DeepSeek
DeepSeek V4-Flashpreview
8.7
AI Panel

Identity & lifecycle

ProviderDeepSeekDeepSeek
Family / tierDeepSeek V3 · LargeDeepSeek V4 · Flash
StatusGApreview
Released2025-08-202026-04-23
Knowledge cutoff2025-062026-02

Architecture & context

Context window128K1M
Max output tokens8K384K
Input modalitiestexttext
Reasoning modeoptionaloptional
Open weightsYesYes

Pricing (per Mtok)

Input$0.21$0.14
Output$0.79$0.28
Cached input$0.021$0.0028
Batch input
Free tierYesYes

Speed

Output speed (tok/s)
Time to first token (s)
Latency tiermediumfast

Trust & deployment

Trains on inputsYesYes
LicenseMITMIT
Clouds
Compliance

AI Panel scoring

Unified score7.78.7
Decision Maker7.58.5
Domain Strategist7.58.5
Finance Lead8.59.7
Domain Practitioner88.5
Power User7.58
Skeptic7.57.5
Value score8.89.9

Benchmarks

MMLU-Pro83.786.2
GPQA Diamond7488.1
AIME 202588
SWE-bench Verified6679
LiveCodeBench8091.6
Aider Polyglot76.3
MRCR Long Context78.7
SimpleQA34.1
Humanity's Last Exam34.8

Scores link to their sources. Missing cells mean the vendor hasn't disclosed a result — honesty over padding.