DeepSeek V3.2 Thinking vs Gemini 3.1 Flash Lite Preview

Change models
IndexAll categories, out of 100.
49.0 ±8.9
49.2 ±6.9 best
GapIndex points below the top.
−0.2 · no clear gap
Top
CoverageIndex weight that has results.
65%
80% best
SpeedOutput tokens per second.
—
116/s
EloLMArena rating and rank.
—
1416
ReleasedMonth of release.
—
2026-03
Input / 1MUS dollars per 1M input tokens.
—
$0.25
Output / 1MUS dollars per 1M output tokens.
—
$1.50
1K chats2K tokens in, 500 out, each.
—
$1.25
1K documents20K tokens in, 1K out, each.
—
$6.50
1K agent tasks200K tokens in, 20K out, each.
—
$80.00
ContextMaximum tokens in one request.
—
1.05M
AgenticMulti-step tasks with tools.
47.4 best
41.1
CodingCode writing and repair.
48.8 best
48.0
ReasoningLogic problems and puzzles.
—
—
MultimodalTasks with images and text.
—
61.8
KnowledgeFacts and expert knowledge.
50.8
52.0 best
MultilingualTasks in many languages.
—
57.1
InstructionTasks with strict prompt rules.
—
—
MathMath problems.
51.9 best
51.3

Benchmarks

12 shared
Only DeepSeek V3.2 Thinking: 4Only Gemini 3.1 Flash Lite Preview: 19
DeepSeek V3.2 Thinking
Gemini 3.1 Flash Lite Preview
76.1%112 / 149
83.8% best47 / 149
TaxEval v2Finance
68.2%100 / 143
71.8% best72 / 143
80.7% best65 / 142
80.1%70 / 142
80.3%66 / 135
81.1% best64 / 135
MMLU ProKnowledge
84.9%59 / 135
86.2% best45 / 135
CorpFin v2Finance
51.0%105 / 131
59.4% best78 / 131
5.1% best116 / 130
0.0%127 / 130
AIMEMath
84.6% best39 / 95
83.3%43 / 95
SWE-benchCoding
67.6% best69 / 87
62.8%74 / 87
36.0% best41 / 66
24.7%52 / 66
55.4% best34 / 53
55.0%35 / 53
2.3%49 / 51
46.1% best33 / 51