Model

DeepSeek-V4-Flash (Think Max)

OPEN

DeepSeek · DeepSeek-V4 family · released Apr 24, 2026

Smaller, cheaper sibling MoE released alongside V4-Pro on the same day; fast default mode. (maximum reasoning-effort tier / 'Pro-Max' mode, DeepSeek's most capable released configuration).

ReasoningCodingVisionFunction callingTool useAgentic

2330.6

Elo · rank #32

Parameters

284B

Active params

13B (MoE)

Context

1M tokens

Architecture

Mixture-of-Experts (284B total / 13B active), hybrid Compressed Sparse Attention + Heavily Compressed Attention

License

MIT

Languages

—

API price (in/out)

$0.14 / $0.28

Modalities

text

Benchmark results

Bar shows position within the tracked field; marker = field best

BrowseCompAgents73.2%#22

best: Kimi K3 · 91.2%

CodeforcesCoding3052#2

best: DeepSeek-V4-Pro (Think Max) · 3206

GDPval-AAAgents1395#20

best: Claude Fable 5 · 1932

GPQA DiamondReasoning88.1%#36

best: GPT-5.6 · 94.6%

Humanity's Last ExamReasoning34.8%#34

best: Claude Sonnet 5 · 57.4%

LiveCodeBenchCoding91.6%#2

best: DeepSeek-V4-Pro (Think Max) · 93.5%

MMLU-ProKnowledge86.2%#21

best: Claude Fable 5 · 91.5%

SWE-bench ProCoding52.6%#31

best: Claude Fable 5 · 80.0%

SWE-bench VerifiedCoding79.0%#21

best: Claude Fable 5 · 95.0%

Terminal-Bench 2.0Coding56.9%#37

best: GPT-5.6 · 88.8%

Run it locally

VRAM @ Q4

—

VRAM @ FP16

—

Fits on (Q4)

Multi-node cluster required

Throughput data unavailable.

Quantizations

—

DeepSeek-V4 family

Elo progression across releases

DeepSeek-V4-Flash (Non-Think)Apr 20261623.2 DeepSeek-V4-Flash (Think High)Apr 20262239.5 DeepSeek-V4-Flash (Think Max)Apr 20262330.6 DeepSeek-V4-Pro (Non-Think)Apr 20261670.4 DeepSeek-V4-Pro (Think High)Apr 20262395.1 DeepSeek-V4-Pro (Think Max)Apr 20262517.1

Resources

HFdeepseek-ai/DeepSeek-V4-Flash DOCShttps://api-docs.deepseek.com/news/news260424/APIAvailable via DeepSeek API

API price $0.14/$0.28 · each benchmark row carries its own source badge (see methodology)