PLPramaana

DeepSeek

DeepSeek V4 Pro

Release date: 4/23/2026

A cost-efficient open reasoning model for mathematics, code, and agentic workflows.

Pramaana IndexAccuracy: StandardLatency: StrongCost: Leading
Accuracy / Pramaana Index
62.48%± 0.82
Latency / Pramaana Index
344.30s
Cost / Test / Pramaana Index
$0.24
Context window
256k
Max output tokens
64k
Input modality
Text
Hyperparameter settings

temperature: 0.2

reasoning: extended

fallbacks: disabled

Benchmark performance for DeepSeek V4 Pro

BenchmarksAccuracyRanking
Pramaana Index62.48% ± 0.8211 / 35
Pramaana Multimodal Index60.47% ± 0.9911 / 24
Legal Research Fidelity56.88% ± 1.1611 / 18
Financial Modeling Agent52.59% ± 1.3311 / 22
Clinical Documentation Safety56.18% ± 1.5011 / 19
MATH-50074.98% ± 1.677 / 112
GPQA Diamond74.29% ± 0.4811 / 121
MMLU-Pro71.78% ± 0.6511 / 119
Adaptive Tutoring Dialogue58.08% ± 0.8211 / 16
SWE-bench Verified54.09% ± 0.999 / 69
LiveCodeBench73.07% ± 1.165 / 127
Agentic Reliability48.67% ± 1.3311 / 12
Strategic Games Arena54.69% ± 1.5011 / 20
Benefits Navigator57.99% ± 1.6711 / 16

License: MIT · Evaluation settings, model endpoints, and benchmark versions are frozen with each published run.

Read our methodology