Claude Haiku 4.5 Thinking vs DeepSeek V3.2
Change modelsClaude Haiku 4.5 ThinkingAnthropic
DeepSeek V3.2DeepSeek
IndexAll categories, out of 100.
43.5 ±7.0
44.2 ±4.9 best
GapIndex points below the top.
−0.7 · no clear gap
Top
CoverageIndex weight that has results.
75%
85% best
SpeedOutput tokens per second.
—
19/s
EloLMArena rating and rank.
—
1425
ReleasedMonth of release.
2025-10
2025-12
Input / 1MUS dollars per 1M input tokens.
—
$0.26
Output / 1MUS dollars per 1M output tokens.
—
$0.38
1K chats2K tokens in, 500 out, each.
—
$0.71
1K documents20K tokens in, 1K out, each.
—
$5.58
1K agent tasks200K tokens in, 20K out, each.
—
$59.60
ContextMaximum tokens in one request.
200K best
164K
AgenticMulti-step tasks with tools.
50.2 best
43.1
CodingCode writing and repair.
42.5
48.8 best
ReasoningLogic problems and puzzles.
—
42.9
MultimodalTasks with images and text.
30.1
—
KnowledgeFacts and expert knowledge.
43.6 best
43.2
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
—
34.8
MathMath problems.
51.6 best
46.8
Benchmarks
12 sharedOnly Claude Haiku 4.5 Thinking: 16Only DeepSeek V3.2: 26
Claude Haiku 4.5 Thinking
DeepSeek V3.2