readable AI benchmarks (simplified)

readable AI benchmarks

DeepSeek

DeepSeek V4 Flash (high)

reasoning · open weights · Apr 24, 2026

DeepSeek V4 Flash (high) has a Quality Score of 41.3, ranking 44th among 53 scored models. Per 1M tokens, pricing is $0.140 input, $0.280 output, and $0.028 cached input. Its Reliability Score is 38.9, ranking 45th among 53 scored models, below average (total average: 50.1). Its Value Score is 42.0, around average compared with the total Value average of 39.6.

Quality Score41.3coverage 91.2%: missing Epoch General ECI, CursorBench 3.2 and DeepSWE v1.1
Value Score42.0
Reliability38.9
Cache Discount80%

Model specification

Reasoning
reasoning
Input modalities
none
Output modalities
none
Context window
not available
Weights
open weights

Token prices USD per 1M tokens

Input
$0.140
Output
$0.280
Cached Input
$0.028
Cache write
not available
Cost per task
$0.047

Capability

Intelligence
37.5
Coding
52.0
Agentic
28.2
Omniscience
-22.3
Correct
35.5%
Blended price
$0.076

Answer outcomes

Attempt rate: 93.3%

Correct 35.5%Incorrect 57.8%Abstained 6.7%

Artificial Analysis benchmarks

GDPval-AA v2
32%
τ³-Banking
20%
Terminal-Bench v2.1
57%
SciCode
42%
Humanity’s Last Exam
28%
GPQA Diamond
87%
CritPt
3%
AA-Omniscience
39%
AA-LCR
63%

Similar models