readable AI benchmarks

DeepSeek

DeepSeek V4 Flash Vision (Reasoning, Max Effort)

reasoning · closed weights · Aug 21, 2026

DeepSeek V4 Flash Vision (Reasoning, Max Effort) has a Quality Score of 53.7, ranking 42nd among 259 scored models. Per 1M tokens, pricing is $0.440 input, $1.32 output, and $0.014 cached input. It accepts text and image input, outputs text, and has a 1M token context window. Its Reliability Score is 41.2, ranking 83rd among 259 scored models, above average (total average: 33.1). Its Value Score is 49.8, high compared with the total Quality average of 26.6.

Quality Score53.7
Value Score49.8
Reliability41.2
Cache Discount97%

Model specification

Reasoning
reasoning
Input modalities
text, image
Output modalities
text
Context window
1M tokens
Weights
closed weights

Token prices USD per 1M tokens

Input
$0.440
Output
$1.32
Cached Input
$0.014
Cached Output
not available
Cost per task
$0.116

Capability

Intelligence
51.5
Coding
65.0
Agentic
52.9
Omniscience
-17.6
Correct
38.6%
Blended price
$0.230

Answer outcomes

Correct 38.6%Incorrect 56.2%Abstained 5.2%

Artificial Analysis benchmarks

GDPval-AA v2
59%
τ³-Banking
41%
Terminal-Bench v2.1
74%
SciCode
47%
Humanity’s Last Exam
34%
GPQA Diamond
91%
CritPt
11%
AA-Omniscience
41%
AA-LCR
78%

Similar models