Community

Platform

Docs Roadmap Tools Migrate Pricing Model Pricing Calculator Works With

Support

Contribute Database monitor Status Contact

Company

About Announcements Changelog Terms Privacy

Community

Platform

Docs Roadmap Tools Migrate Pricing Model Pricing Calculator Works With

Support

Contribute Database monitor Status Contact

Company

About Announcements Changelog Terms Privacy

© 2025 AI StatsIf you run into any issues or notice any data errors, please visit our GitHub and report an issue.

Quickstart

Start calling this model with endpoint-specific examples.

Benchmarks

Headline benchmark standings and comparison context.

About

Key dates, capabilities, and model metadata.

Models Providers Apps Rankings Playground

Models Providers Apps Rankings Playground

DeepSeek V4 Pro Benchmarks - Performance Highlights | AI Stats

DeepSeek V4 Pro

DeepSeek V4 Pro

DeepSeek

Overview Playground Providers Pricing Performance Apps Activity Quickstart Benchmarks Family Timeline

Benchmarks

Headline benchmark standings and comparison context.

Highlights

Top benchmark results for deepseek/deepseek-v4-pro.

83.10-

38.30%-

90.20%-

87.50%-

59.20%-

83.40%-

93.10%-

Chinese SimpleQA

84.40%-

85.20%-

90.90%-

90.80-

3206-

62%-

88.70%-

FACTS Parametric

62.60%-

1554-

90.10%-

92.60%-

88%-

95.20%-

76.80%-

Humanity's Last Exam

48.20%-

IMO Answer Bench

89.80%-

93.50%-

51.50%-

64.50%-

84.40%-

90.10%-

90.80%-

87.50%-

90.30%-

83.50%-

51.10%-

Scale MCP Atlas

74.20%-

SimpleQA Verified

57.90-

53.90%-

SWE Bench Multilingual

76.20%-

55.40%-

80.60%-

Terminal Bench 2.0

67.90%-

51.80%-

85.60%-

81.50%-

Detailed benchmark comparisons now live in the Compare tool.

Open Compare Tool Browse Comparisons