GeekRadar / The comparison desk

COMPARE.

AI models

anthropic

Claude Haiku 3.5

AI model Released Oct 22, 2024

Context window: 200,000 tokens
Maximum output: 8,192 tokens
Open weights: No

Capability & preference

Human preference · Arena

Overall text preferences from head-to-head votes. Not factual accuracy or a task-success rate.

Claude Haiku 3.5 1255.4 claude-3-5-haiku-20241022 · 95% interval 1252.2–1258.6 · 69,061 votes · 2026-10-15

Model performance

Radar task score: four tests, equal weight. An experimental summary of these tasks, not a universal intelligence rating.

Claude Haiku 3.5

Tested configuration Radar task score

Unrated

1/4 core tests · not enough evidence

1 benchmark results for this configuration

Scientific reasoning

GPQA Diamond · 25%

Expert-written science questions; strict multiple-choice scoring. Chance accuracy is 25%.

Claude Haiku 3.5 38.1^*% Approx. 95% interval: 33.0–43.3% · 2025-03-12 · Older result

Novel problem solving

Mystery Game Puzzles · 25%

Best-next-move puzzles in an undisclosed game variant. A narrow reasoning test, not general intelligence.

Claude Haiku 3.5 Not available

Advanced mathematics

FrontierMath Tiers 1-3 · 25%

Private v2 mathematics problems with Python tools. Earlier versions and Tier 4 are excluded.

Claude Haiku 3.5 Not available

Factual answers

SimpleQA Verified · 25%

Correct factual answers under a guessing prompt, graded by a model. Not a hallucination or abstention score.

Claude Haiku 3.5 Not available

Bars show reported task scores. Whiskers are approximate 95% intervals where standard errors are available. Different reasoning budgets are not equal-cost tests. Missing results are never zero.

Recorded results

Only the same benchmark, version, workload, scenario and unit share a row. Missing results are not zero. System results include the full submitted hardware and software configuration.

Arena text preference / chinesehuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,228.72 rating

Confidence interval: 1,219.41 – 1,238.02

4659 votes

Oct 15, 2026

Category: chinese
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / codinghuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,286.18 rating

Confidence interval: 1,280.23 – 1,292.14

11143 votes

Oct 15, 2026

Category: coding
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / creative_writinghuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,232.44 rating

Confidence interval: 1,226.01 – 1,238.88

9836 votes

Oct 15, 2026

Category: creative_writing
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / englishhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,267.48 rating

Confidence interval: 1,263.43 – 1,271.53

35506 votes

Oct 15, 2026

Category: english
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / exclude_tieshuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,171.09 rating

Confidence interval: 1,166.43 – 1,175.75

48599 votes

Oct 15, 2026

Category: exclude_ties
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / experthuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,207.62 rating

Confidence interval: 1,197.10 – 1,218.14

3526 votes

Oct 15, 2026

Category: expert
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / frenchhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,263.52 rating

Confidence interval: 1,241.59 – 1,285.46

838 votes

Oct 15, 2026

Category: french
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / germanhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,236.81 rating

Confidence interval: 1,221.79 – 1,251.84

1739 votes

Oct 15, 2026

Category: german
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / hard_promptshuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,250.82 rating

Confidence interval: 1,246.23 – 1,255.41

23545 votes

Oct 15, 2026

Category: hard_prompts
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / hard_prompts_englishhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,264.21 rating

Confidence interval: 1,258.39 – 1,270.04

12328 votes

Oct 15, 2026

Category: hard_prompts_english
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,222.61 rating

Confidence interval: 1,215.77 – 1,229.46

9073 votes

Oct 15, 2026

Category: industry_business_and_management_and_financial_operations
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,223.94 rating

Confidence interval: 1,218.14 – 1,229.73

12382 votes

Oct 15, 2026

Category: industry_entertainment_and_sports_and_media
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / industry_legal_and_governmenthuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,258.43 rating

Confidence interval: 1,248.65 – 1,268.21

4229 votes

Oct 15, 2026

Category: industry_legal_and_government
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,246.99 rating

Confidence interval: 1,240.82 – 1,253.16

11848 votes

Oct 15, 2026

Category: industry_life_and_physical_and_social_science
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / industry_mathematicalhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,248.83 rating

Confidence interval: 1,240.85 – 1,256.82

5757 votes

Oct 15, 2026

Category: industry_mathematical
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,227.41 rating

Confidence interval: 1,216.54 – 1,238.29

3640 votes

Oct 15, 2026

Category: industry_medicine_and_healthcare
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,272.29 rating

Confidence interval: 1,267.37 – 1,277.22

19602 votes

Oct 15, 2026

Category: industry_software_and_it_services
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,256.38 rating

Confidence interval: 1,251.23 – 1,261.52

16811 votes

Oct 15, 2026

Category: industry_writing_and_literature_and_language
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / instruction_followinghuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,240.49 rating

Confidence interval: 1,235.84 – 1,245.14

21925 votes

Oct 15, 2026

Category: instruction_following
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / japanesehuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,174.96 rating

Confidence interval: 1,158.69 – 1,191.24

1546 votes

Oct 15, 2026

Category: japanese
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / koreanhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,173.09 rating

Confidence interval: 1,154.48 – 1,191.71

1211 votes

Oct 15, 2026

Category: korean
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / longer_queryhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,260.35 rating

Confidence interval: 1,254.29 – 1,266.41

10962 votes

Oct 15, 2026

Category: longer_query
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / mathhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,243.83 rating

Confidence interval: 1,236.32 – 1,251.34

6334 votes

Oct 15, 2026

Category: math
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / multi_turnhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,265.07 rating

Confidence interval: 1,258.86 – 1,271.28

11176 votes

Oct 15, 2026

Category: multi_turn
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / non_englishhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,237.85 rating

Confidence interval: 1,233.76 – 1,241.93

33547 votes

Oct 15, 2026

Category: non_english
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / overallhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,255.38 rating

Confidence interval: 1,252.15 – 1,258.61

69061 votes

Oct 15, 2026

Category: overall
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / polishhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,249.47 rating

Confidence interval: 1,236.80 – 1,262.14

2717 votes

Oct 15, 2026

Category: polish
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / russianhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,252.78 rating

Confidence interval: 1,245.33 – 1,260.22

6652 votes

Oct 15, 2026

Category: russian
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Arena text preference / spanishhuman preference · 2026-10-15 · rating

Claude Haiku 3.5

1,261.18 rating

Confidence interval: 1,241.22 – 1,281.14

963 votes

Oct 15, 2026

Category: spanish
Evaluated configuration: claude-3-5-haiku-20241022
Evaluation type: human preference

Run it locally

Inference · memory estimate

Claude Haiku 3.5

Closed weights. No supported local deployment or GPU requirement is published.

API cost

USD / month

Input tokens / monthOutput tokens / month

Standard text-token estimate. Excludes taxes, caching, batch discounts, reasoning and multimodal surcharges. Provider limits and regional availability may differ.

Claude Haiku 3.5

Provider offer

Capabilities

Context window: 200,000 tokens
Maximum output: 8,192 tokens
Reasoning: No
Tool calling: Yes
Open weights: No
Knowledge cutoff: 2024-07-31
Model license: Proprietary

Input modalities

text
image
pdf

Output modalities

text

* Provisional data