Inference
EN
Anthropic

Claude Haiku 4.5 Thinking

Claude Haiku 4.5 Thinking is a reasoning model from Anthropic in the Claude Haiku 4.5 family. 13 benchmarks count toward its index, in 5 categories.

  • Closed weights
  • Reasoning
  • Input: text
  • Released 2025-10
RankedEnough results for an index.
IndexAll categories, out of 100.43.5±7.0Median 51.0
CoverageIndex weight that has results.75%
SpeedOutput tokens per second.—Median 67/s
First tokenSeconds to the first output token.—Median 1.3 s
Input / 1MUS dollars per 1M input tokens.—Median $0.60
Output / 1MUS dollars per 1M output tokens.—Median $2.55
ContextMaximum tokens in one request.200KMedian 262K
EloLMArena rating and rank.—Median 1422

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.3 counted50.2128 / 218
CodingCode writing and repair.5 counted42.5175 / 209
ReasoningLogic problems and puzzles.None counted——
MultimodalTasks with images and text.2 counted30.181 / 84
KnowledgeFacts and expert knowledge.2 counted43.6175 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.None counted——
MathMath problems.1 counted51.6—

Results

13 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
Terminal-Bench 1.0Agentic50.0% ±5.610 / 47
AIMEMath82.7% ±0.944 / 95
Terminal-Bench 2.0 · Vals AIAgentic38.2% ±5.236 / 66
GPQA Diamond · Vals AIKnowledge72.2% ±2.588 / 135
IOI v1Coding6.2% ±2.543 / 61
MMLU ProKnowledge78.7% ±0.5100 / 135
SWE-benchCoding66.6% ±2.170 / 87
ParseBench: document parsing for AI agents (LlamaIndex leaderboard)Multimodal51.2%33 / 39
Vibe Code Bench v1.1Coding11.4% ±3.1105 / 123
Code MigrationCoding7.6% ±1.166 / 77
Terminal-Bench 2.1 · Vals AIAgentic43.8% ±1.965 / 75
LiveCodeBenchCoding41.2% ±1.1129 / 142
MMMU ProMultimodal46.1% ±1.290 / 92
MGSMMultilingual92.1% ±1.621 / 74
MedScribeHealthcare85.2% ±1.931 / 109
CorpFin v2Finance60.6% ±1.066 / 131
ProgramBenchCoding0.0%34 / 65
LegalBenchLegal81.2% ±0.582 / 149
CaseLaw v2Legal56.5% ±0.930 / 53
MortgageTaxFinance62.2% ±1.056 / 97
Finance Agent v1.1Finance46.9% ±2.832 / 51
TaxEval v2Finance67.5% ±0.9106 / 143
MedQAHealthcare79.6% ±0.474 / 95
MedCodeHealthcare32.7% ±2.087 / 106
Legal Research BenchLegal10.6% ±2.179 / 90
Finance Agent (v2)Finance31.0% ±0.670 / 79
EMBFinance22.7% ±2.469 / 75
Tax Agent BenchFinance5.9% ±1.577 / 82

Sources

BenchLMCC BY-NC 4.0Vals AINo licence stated

History

5 releases
Index44.6 → 43.5
23 Oct 202612 Nov 2026

Questions

How good is Claude Haiku 4.5 Thinking?
Its index is 43.5 out of 100. That is place 187 of 283 ranked models.
What is Claude Haiku 4.5 Thinking best at?
Agentic: 50.2 out of 100, place 128 of 218.
What is the context window of Claude Haiku 4.5 Thinking?
200K tokens.
Is Claude Haiku 4.5 Thinking open source?
No. Its weights are not published.
When was Claude Haiku 4.5 Thinking released?
October 2025.

Badge

Claude Haiku 4.5 Thinking on Inference360: place 187 of 283, index 43.5
HTML
<a href="https://inference360.ai/models/claude-haiku-4-5-thinking"><img src="https://inference360.ai/badges/claude-haiku-4-5-thinking.svg" alt="Claude Haiku 4.5 Thinking on Inference360: place 187 of 283, index 43.5" height="20"></a>
Markdown
[![Claude Haiku 4.5 Thinking on Inference360: place 187 of 283, index 43.5](https://inference360.ai/badges/claude-haiku-4-5-thinking.svg)](https://inference360.ai/models/claude-haiku-4-5-thinking)

Compare with

Trinity-Large-Thinking43.5Claude Opus 443.9DeepSeek V3.244.2Claude Sonnet 4.547.9

More from Anthropic

All models
Claude Haiku 4.543.2Claude Sonnet 4.5 Thinking53.8Claude Sonnet 4.547.9Claude Opus 4.5 Thinking56.3Claude Opus 4.552.0