DeepSeek V4 Pro 0423 vs Gemini 3 Flash Preview

Change models
IndexAll categories, out of 100.
57.6 ±4.8 best
57.5 ±5.3
GapIndex points below the top.
Top
−0.1 · no clear gap
CoverageIndex weight that has results.
85%
90% best
SpeedOutput tokens per second.
89/s best
78/s
EloLMArena rating and rank.
1451
—
ReleasedMonth of release.
2026-04
2025-12
Input / 1MUS dollars per 1M input tokens.
$0.892
$0.50 best
Output / 1MUS dollars per 1M output tokens.
$1.78 best
$3.00
Batch output / 1MBatch price per 1M output tokens.
—
$1.50
1K chats2K tokens in, 500 out, each.
$2.68
$2.50 best
1K documents20K tokens in, 1K out, each.
$19.63
$13.00 best
1K agent tasks200K tokens in, 20K out, each.
$214.15
$160.00 best
ContextMaximum tokens in one request.
1.05M
1.05M
AgenticMulti-step tasks with tools.
55.9
57.3 best
CodingCode writing and repair.
59.2 best
55.0
ReasoningLogic problems and puzzles.
54.6
60.3 best
MultimodalTasks with images and text.
—
69.2
KnowledgeFacts and expert knowledge.
59.9 best
57.0
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
52.7
—
MathMath problems.
57.6 best
57.3

Benchmarks

21 shared
Only DeepSeek V4 Pro 0423: 20Only Gemini 3 Flash Preview: 11
DeepSeek V4 Pro 0423
Gemini 3 Flash Preview
80.3%87 / 149
86.9% best11 / 149
TaxEval v2Finance
72.1%68 / 143
73.9% best45 / 143
87.5% best13 / 142
85.6%29 / 142
89.4% best30 / 135
87.9%37 / 135
MMLU ProKnowledge
87.2%32 / 135
88.6% best18 / 135
CorpFin v2Finance
61.4%52 / 131
66.4% best19 / 131
49.9% best66 / 123
20.2%90 / 123
MedScribeHealthcare
75.1% best81 / 109
69.9%95 / 109
MedCodeHealthcare
40.5%64 / 106
55.9% best5 / 106
23.1% best66 / 90
18.3%68 / 90
SWE-benchCoding
77.4% best35 / 87
75.0%44 / 87
25.2% best44 / 82
11.8%64 / 82
44.1% best58 / 79
42.6%60 / 79
26.2% best46 / 77
6.4%69 / 77
EMBFinance
51.6% best55 / 75
26.0%67 / 75
50.2%58 / 75
53.9% best53 / 75
56.2% best14 / 66
51.7%18 / 66
0.0%34 / 65
0.0%34 / 65
IOI v1Coding
35.8%15 / 61
39.1% best13 / 61
59.4% best27 / 53
55.8%32 / 53