AI Benchmark Model
DeepSeek V4 Flash (Reasoning, Max Effort)
Benchmark scores, pricing, speed, and model comparisons for DeepSeek V4 Flash (Reasoning, Max Effort).
Key scores
DeepSeek V4 Flash, released by DeepSeek on April 24, 2026, is a reasoning-focused model designed for high-performance tasks. The model achieves an intelligence index of 40.3 and a coding index of 56.2. Performance metrics indicate an output speed of 111.163 tokens per second with a time to first token of 0.922 seconds. Benchmark results include 0.894 on GPQA, 0.321 on HLE, 0.449 on SciCode, 0.7918 on IFBench, 0.63 on LCR, 0.3561 on TerminalBench Hard, and 0.9503 on TAU2. Pricing for the model is set at $0.14 per million tokens for input and $0.28 per million tokens for output, resulting in a blended cost of $0.17 per million tokens.
Benchmark results
7 tracked benchmark rows for DeepSeek V4 Flash (Reasoning, Max Effort).
TAU2 Other
95.0% GPQA Reasoning
89.4% IFBench Other
79.2% LCR Other
63.0% SciCode Other
44.9% TerminalBench Hard Other
35.6% HLE Other
32.1%
Comments (0)
to join the discussion
No comments yet
Be the first to share your thoughts!