BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases9
AnthropicCompare with another model

Claude 3.5 Haiku

Claude 3.5 Haiku is an Anthropic proprietary model, released 4 Nov 2024. Its best configuration ranks #675 of 785 on the BenchLeader Index at 39.7 ±3.2, in the lower half. The ± is the point: 239 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in human preference (48) and lowest in multimodal (33). Last measured 22 Oct 2026.

Blended price
–
Output speed
–
First answer
–
Context
200k
Full answer
–
Cost per run
–
Released
4 Nov 2024
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Overall index40
  2. Reasoning38
  3. Coding35
  4. Agents & tools36
  5. Maths37
  6. Knowledge37
  7. Instruction following45
  8. Human preference48
  9. Multimodal33
  10. Long context38
  11. Composite39

Versions

Anthropic has shipped 6 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.

ModelReleasedIndexRank
Claude Haiku 5.5xhigh7 Oct 202661.5#95
Claude Haiku 4.5no reasoning15 Oct 202548.5#424
Claude 3.5 Haikuthis page4 Nov 202439.7#675
Claude 3 Haiku7 Mar 202437.0#738
Claude Haiku 3–––
Claude Haiku 4thinking–––

How its index has moved

8 Sept 2026 to 18 Oct 2026
374145

The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.

Benchmark results

See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

Benchmarknot statedSourceTrend
GPQA Diamond38.1%#252Epoch AI Benchmarking Hub
Arena Hard Prompts1345#219Arena (LMArena)
GPQA Diamond (AA)not in index40.8%#476Artificial Analysis
Humanity's Last Exam (AA)not in index3.6%#581Artificial Analysis
GPQA Diamond (Vals)not in index37.9%#124Vals AI
GPQA Diamond (HELM)not in index36.3%#55HELM Capabilities
CritPt0.0%#315Artificial Analysis
DTBenchnot in index56.7%#193Epoch AI Benchmarking Hub

Coding

Benchmarknot statedSourceTrend
SciCode27.4%#203SciCode
WeirdML30.7%#132WeirdML
Arena Coding1386#209Arena (LMArena)
LiveCodeBench41.9%#122Vals AI
Aider Polyglot28.0%#36Aider polyglot leaderboard
CadEvalnot in index32.0%#10Epoch AI Benchmarking Hub
Terminal-Bench 2.1 (AA)not in index10.1%#194Artificial Analysis

Agents & tools

Benchmarknot statedSourceTrend
Terminal-Bench Hard2.3%#320Artificial Analysis
τ²-Bench Telecom (AA)not in index24.6%#300Artificial Analysis
BALROGnot in index19.3%#25BALROG
BrowseCompnot in index0.3%#36OpenAI

Maths

Benchmarknot statedSourceTrend
OTIS Mock AIME4.3%#249Epoch AI Benchmarking Hub
MATH Level 546.4%#57Epoch AI Benchmarking Hub
AIME (Vals)3.3%#90Vals AI
MGSM84.6%#65Vals AI
Omni-MATH (HELM)22.4%#54HELM Capabilities
MATH 500not in index64.2%#54Vals AI
Arena Maths1287#241Arena (LMArena)

Knowledge

Benchmarknot statedSourceTrend
AA-Omniscience-22.5#254Artificial Analysis
MMLU-Pro64.1%#124Vals AI
LegalBench70.3%#118Vals AI
CorpFin50.8%#100Vals AI
TaxEval57.4%#130Vals AI
MMLU-Pro (HELM)not in index60.5%#46HELM Capabilities
MMLUnot in index74.3%#40Epoch AI Benchmarking Hub
AA-Omniscience: accuracynot in index13.2%#461Artificial Analysis
AA-Omniscience: non-hallucinationnot in index58.9%#83Artificial Analysis

Instruction following

Benchmarknot statedSourceTrend
IFBench42.8%#234Artificial Analysis
IFEval (HELM)79.2%#50HELM Capabilities
WildBench (HELM)76.0%#49HELM Capabilities
Arena Instruction Followingnot in index1316#219Arena (LMArena)

Human preference

Benchmarknot statedSourceTrend
Arena Text1325#228Arena (LMArena)
Arena Creative Writingnot in index1305#209Arena (LMArena)
Arena Multi-turnnot in index1328#213Arena (LMArena)
Arena Longer Queriesnot in index1346#206Arena (LMArena)
Lech Mazur Creative Writingnot in index73.5%#28Lech Mazur

Multimodal

Benchmarknot statedSourceTrend
Arena Vision1093#127Arena (LMArena)
MMMU-Pro45.6%#253Artificial Analysis
GeoBenchnot in index34.0%#27GeoBench

Long context

Benchmarknot statedSourceTrend
AA-LCR27.3%#405Artificial Analysis

Composite

Benchmarknot statedSourceTrend
Epoch Capabilities Indexnot in index127.1#157Epoch AI Benchmarking Hub
AA Intelligence Index v4.3.28.9#419Artificial Analysis
HELM Capabilities meannot in index54.9%#54HELM Capabilities

Safety & honesty

Benchmarknot statedSourceTrend
FORTRESSnot in index30.4%#22Scale AI SEAL

More from Anthropic

New this month

See also

Best for codingAll Anthropic models

Questions people ask

What are Claude 3.5 Haiku's benchmark scores?
As of 25 Oct 2026, Claude 3.5 Haiku scores 39.7 ±3.2 on the BenchLeader Index, strongest in human preference (48) and weakest in multimodal (33). That rests on 28 independent benchmark results, and every score on this page links to its publisher.
How does Claude 3.5 Haiku rank?
Claude 3.5 Haiku is #675 of 785 ranked configurations on 25 Oct 2026. 239 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
When was Claude 3.5 Haiku released?
Anthropic released Claude 3.5 Haiku on 4 Nov 2024.

Data as of 25 Oct 2026. Compare with another model.

Cite as: BenchLeader, “Claude 3.5 Haiku: benchmarks, pricing, speed and rank”, data as of 25 Oct 2026, https://www.benchleader.com/models/claude-3-5-haiku.