DeepSeek · LIVEBENCH 2026-06-25

DeepSeek V4 Pro 0813

Published benchmark results for this specific model configuration.

deepseek-v4-pro-0813

Overall score77.4 / 100

Cost / successful task$0.044USD · benchmark workload

Weight accessOpen weights

Where this configuration performs

Category averages from the same release, on a 0–100 scale.

Reasoning85.8

Coding77.2

Agentic Coding54.9

Mathematics95.1

Data Analysis79.2

Language82.1

Instruction Following67.7

Task-level results

theory of mind84.6zebra puzzle96.8spatial96.0logic with navigation66.0code generation76.1code completion78.3javascript68.2typescript46.7python50.0AMPS Hard98.0integrals with game94.0math comp97.1olympiad91.3consecutive events90.2tablejoin49.5tablereformat98.0connections100.0plot unscrambling64.2typos82.0paraphrase68.6simplify61.6story generation74.9summarize65.7

Source files: Scores ↗ Categories ↗ Costs ↗ . Missing cost is unknown, never free. Compare configurations within the same release.