AI Benchmark Model
DeepSeek V4 Pro (Non-reasoning)
Benchmark scores, pricing, speed, and model comparisons for DeepSeek V4 Pro (Non-reasoning).
Key scores
DeepSeek V4 Pro (Non-reasoning) is an artificial intelligence model released by DeepSeek on April 24, 2026. The model holds an intelligence index of 31.9. In terms of performance metrics, it achieves an output speed of 62.894 tokens per second with a time to first token of 1.24 seconds. Pricing for the model is set at $0.43 per million tokens for input and $0.87 per million tokens for output, resulting in a blended rate of $0.54 per million tokens. Benchmark results for the model include 0.717 on GPQA, 0.082 on HLE, 0.424 on SciCode, 0.458 on IFBench, 0.497 on LCR, 0.364 on TerminalBench Hard, and 0.912 on TAU2.
Benchmark results
7 tracked benchmark rows for DeepSeek V4 Pro (Non-reasoning).
TAU2 Other
91.2% GPQA Reasoning
71.7% LCR Other
49.7% IFBench Other
45.8% SciCode Other
42.4% TerminalBench Hard Other
36.4% HLE Other
8.2%
Comments (0)
to join the discussion
No comments yet
Be the first to share your thoughts!