DeepSeek V4 Pro (Reasoning, High Effort)

Benchmark scores, pricing, speed and comparisons for DeepSeek V4 Pro (Reasoning, High Effort).

DeepSeek V4 Pro (Reasoning, High Effort) is an AI model from DeepSeek, released 2026-04-24. It scores 43.7 on the Intelligence Index and 58.7 on the Coding Index.

DeepSeek V4 Pro (Reasoning, High Effort) at a glance

DeepSeek released DeepSeek V4 Pro (Reasoning, High Effort) on April 24, 2026. It is marked as retired in the catalogue, so treat its numbers as a historical record.

DeepSeek V4 Pro (Reasoning, High Effort) scores 43.7 on the Intelligence Index, the overall score the AI Benchmark rankings sort models by. It scores 58.7 on the narrower Coding Index. Of its 7 individual benchmark results, the highest is 94.2% on TAU2 and the lowest is 35.2% on HLE. These scores come from independent third-party model evaluations, which Franklin AI syncs automatically each day.

It costs $0.43 per million input tokens and $0.87 per million output tokens, so a sample workload of 1 million input tokens and 200,000 output tokens costs about $0.61. It writes about 61 output tokens per second, so a 500-token answer takes roughly 8.2 seconds once output starts, after a wait of about 1.37 seconds for the first token.

The benchmark data has no Math Index for DeepSeek V4 Pro (Reasoning, High Effort) yet.

Every pair of models in the catalogue has a head-to-head page. Pick DeepSeek V4 Pro (Reasoning, High Effort) and a rival on the model comparisons page to compare their scores, prices and speed side by side.

Key scores

Intelligence43.7Index score
Coding58.7Index score
Speed61.2Tokens/sec
Blended Cost$0.54Per 1M tokens

How does DeepSeek V4 Pro (Reasoning, High Effort) score on each benchmark?

7 benchmark results for DeepSeek V4 Pro (Reasoning, High Effort).

TAU2 Other
94.2%
GPQA Reasoning
90.5%
IFBench Other
71.3%
LCR Other
67.0%
SciCode Other
46.4%
TerminalBench Hard Other
41.7%
HLE Other
35.2%