Translate
Independent model intelligence

Find the right AI model,
not just the loudest one.

Compare current model capability, context, modalities, public preference and API economics in one continuously refreshed decision surface.

LMArena models.dev OpenRouter Artificial Analysis when available
Back to leaderboard
DeepSeek · Active

deepseek-v4-pro

deepseek-thinking family · launched 2026-04-24

ReasoningToolsStructuredOpen weights
AITopic score46.3#48 overall

Open MoE flagship with million-token context for coding and long agent runs

Evidence coverage3%1 evaluations · 1 institutions
Context / maximum output1M / 16.4ktokens
Cache / input / output$0.1 · $1.3 · $2.6USD per million tokens
Value / freshness63.9 / 65relative value · release freshness
Product profile

What this model supports

Inputtext
Outputtext
Knowledge cutoff2025-05
WeightsOpen weights
Capability profile

Strengths by domain

Coding—0 public results
Research & reasoning—0 public results
Professional work—0 public results
Knowledge & instruction—0 public results
Behind the score

Every result has a source

Raw results and standardized readings are shown together. Missing tests add neither points nor penalties.

Human preference & indices2 results

AA IndexArtificial Analysis46.34 indexReading 46.3

Imported from a directly fetched third-party leaderboard.

Configuration: deepseek-v4-proOpen evidence
Arena TextLMArena1,451.2 EloReading 46.3

Direct LMArena text leaderboard result.

Open evidence
Same builder

More from DeepSeek