DataLearner logo

DeepSeek-V4-Flash Benchmark Details

DeepSeek-V4-Flash currently shows benchmark results led by LiveCodeBench (4 / 123, score 91.60), MMLU Pro (16 / 132, score 86.40), GPQA Diamond (31 / 187, score 88.10).

Benchmark Results

DeepSeek-V4-Flash

Benchmark Results

Thinking
Tool usage

General Knowledge

12 evaluations
Benchmark / mode
Score
Rank/total
GPQA Diamond
Standard Mode
71.20
112 / 187
GPQA Diamond
High
87.40
38 / 187
GPQA Diamond
Max
88.10
31 / 187
MMLU Pro
Standard Mode
83
45 / 132
MMLU Pro
High
86.40
16 / 132
MMLU Pro
Max
86.20
17 / 132
LiveBench
Standard Mode
67.25
49 / 115
HLE
Standard Mode
8.10
155 / 172
HLE
High
29.40
91 / 172
HLE
HighTools
40.30
62 / 172
HLE
Max
34.80
74 / 172
HLE
Thinking Level · Extra HighTools
45.10
42 / 172

Coding and Software Engineer

16 evaluations
Benchmark / mode
Score
Rank/total
CodeForces
High
2816
5 / 16
CodeForces
Max
3052
3 / 16
LiveCodeBench
Standard Mode
55.20
83 / 123
LiveCodeBench
High
88.40
8 / 123
LiveCodeBench
Max
91.60
4 / 123
SWE-bench Verified
Standard ModeTools
73.70
44 / 112
SWE-bench Verified
HighTools
78.60
23 / 112
SWE-bench Verified
Thinking Level · Extra HighTools
79
20 / 112
SWE-bench Multilingual
Standard ModeTools
69.70
19 / 23
SWE-bench Multilingual
HighTools
70.20
17 / 23
SWE-bench Multilingual
Thinking Level · Extra HighTools
73.30
12 / 23
DeepSWE
MaxTools
54.40
9 / 20
NL2Repo-Bench
MaxTools
54.20
1 / 1
SWE-Bench Pro - Public
Standard ModeTools
49.10
44 / 54
SWE-Bench Pro - Public
HighTools
52.30
37 / 54
SWE-Bench Pro - Public
Thinking Level · Extra HighTools
52.60
35 / 54

Common Sense Reasoning

1 evaluations
Benchmark / mode
Score
Rank/total
Simple Bench
Standard Mode
46.30
32 / 63

AI Agent - Information Search

2 evaluations
Benchmark / mode
Score
Rank/total
BrowseComp
HighTools
53.50
40 / 53
BrowseComp
Thinking Level · Extra HighTools
73.20
28 / 53

AI Agent - Tool Usage

7 evaluations
Benchmark / mode
Score
Rank/total
TerminalBench 2.1
MaxTools
82.70
9 / 28
CyberGym
MaxTools
76.70
1 / 1
Toolathlon-Verified
MaxTools
70.30
2 / 2
Terminal Bench 2.0
Standard ModeTools
49.10
35 / 47
Terminal Bench 2.0
HighTools
56.60
27 / 47
Terminal Bench 2.0
Thinking Level · Extra HighTools
56.90
25 / 47
Automation Bench
MaxTools
25.10
3 / 3

Math and Reasoning

3 evaluations
Benchmark / mode
Score
Rank/total
IMO-AnswerBench
Standard Mode
41.90
20 / 21
IMO-AnswerBench
High
85.10
10 / 21
IMO-AnswerBench
Max
88.40
5 / 21

Productivity Knowledge

1 evaluations
Benchmark / mode
Score
Rank/total
GDPval-AA
Thinking Level · Extra HighTools
1395
6 / 21

Agent Level Benchmark

1 evaluations
Benchmark / mode
Score
Rank/total
Agents' Last Exam
MaxTools
25.20
5 / 5
Compare with other models