Claude Haiku 3.5

Anthropic · anthropic/claude-3-5-haiku-20241022

ProviderAnthropic
Familyclaude-haiku
Typevlm
Statusactive
Released2024-10-22
Updated2024-10-22

Reported benchmark scores

These scores carry one collection date for the whole card (2026-04) and a source list rather than a source per score, so they cannot be attributed individually. They are shown as reported and are not verified evidence. Benchmark pages carry the reviewed, dated evidence where it exists.
BenchmarkCatalogue standingScore
Aider Polyglot Benchmarkunassessed28.0
Arena Elo — Codingunassessed1270.0
Arena Elo — Hard Promptsunassessed1343.3
Arena Elo — Mathunassessed1250.0
Arena Elo — Overall (Text)unassessed1260.0
Arena Elo — Style Controlunassessed1322.4
Arena Elo — Visionunassessed1126.8
BBQ (Bias Benchmark for QA)unassessed84.5
GPQA Diamondunassessed51.2
HELM Safetyunassessed89.8
HumanEvalunassessed82.5
IFEvalunassessed80.5
LiveCodeBenchunassessed28.8
MATH-500unassessed69.5
MedQAunassessed77.8
MGSM (Multilingual Grade School Math)unassessed85.6
MMLU-Prounassessed65.3
MultiPL-E: C#unassessed72.5
MultiPL-E: Juliaunassessed51.8
MultiPL-E: Kotlin (unconfirmed)unassessed62.8
MultiPL-E: Luaunassessed46.8
MultiPL-E: Perlunassessed42.5
MultiPL-E: PHPunassessed68.8
MultiPL-E: Runassessed48.5
MultiPL-E: Rubyunassessed58.2
MultiPL-E: Scalaunassessed52.5
MultiPL-E: Swiftunassessed56.5
SWE-bench Verifiedunassessed40.6
ToxiGenunassessed92.5

Data

This card as JSON · Edit on GitHub