DeepSeek V3.2 vs DeepSeek V3.2 (Thinking)
Change modelsDeepSeek V3.2DeepSeek
DeepSeek V3.2 (Thinking)DeepSeek
IndexAll categories, out of 100.
44.2 ±4.9
49.4 ±8.6 best
GapIndex points below the top.
−5.2 · no clear gap
Top
CoverageIndex weight that has results.
85% best
65%
SpeedOutput tokens per second.
19/s
—
EloLMArena rating and rank.
1425 best
1420
ReleasedMonth of release.
2025-12
2025-12
Input / 1MUS dollars per 1M input tokens.
$0.26 best
$0.55
Output / 1MUS dollars per 1M output tokens.
$0.38 best
$2.19
1K chats2K tokens in, 500 out, each.
$0.71 best
$2.20
1K documents20K tokens in, 1K out, each.
$5.58 best
$13.19
1K agent tasks200K tokens in, 20K out, each.
$59.60 best
$153.80
ContextMaximum tokens in one request.
164K best
128K
AgenticMulti-step tasks with tools.
43.1
47.7 best
CodingCode writing and repair.
48.8
49.2 best
ReasoningLogic problems and puzzles.
42.9
—
MultimodalTasks with images and text.
—
—
KnowledgeFacts and expert knowledge.
43.2
51.2 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
34.8
—
MathMath problems.
46.8
51.0 best
Benchmarks
12 sharedOnly DeepSeek V3.2: 26Only DeepSeek V3.2 (Thinking): 5
DeepSeek V3.2
DeepSeek V3.2 (Thinking)