Anthropic·released 2024-03-071 source
Claude 3 Haiku benchmark scores: 10 benchmarks tracked. 30% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 10 benchmarks·best result 65.07 on MMLU (unverified)·3 of 10 independently reproduced·$0.25/$1.25 per M tokens
Consensus: LiteLLM · See every model’s pricing →
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| MMLU | 65.07accuracy (%) | unverified· optimizedT1 | 2024-03-07Stanford CRFM Leaderboard |
| ScienceQA | 62.67accuracy (%) | self-reported· optimizedT1 | 2024-03-07Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone |
| Winogrande | 48.4accuracy (%) | unverified· optimizedT1 | 2024-03-07The Claude 3 Model Family: Opus, Sonnet, Haiku |
| DTBench | 16.88accuracy (%) | unverifiedT1 | 2024-03-07https://conceptualreasoning.ai/dtbench |
| GPQA diamond | 15.07accuracy (%) | reproduced· optimizedT1 | 2024-03-07Epoch AI |
| MATH level 5 | 14.88accuracy (%) | reproduced· optimizedT1 | 2024-03-07Epoch AI |
| CadEval | 12accuracy (%) | unverifiedT1 | 2024-03-07CadEval Dashboard |
| LMCA | 10.4accuracy (%) | unverifiedT1 | 2024-03-07https://conceptualreasoning.ai/lmca |
| WeirdML | 9.84accuracy (%) | unverifiedT1 | 2024-03-07https://htihle.github.io/weirdml.html |
| OTIS Mock AIME 2024-2025 | 1.71accuracy (%) | reproduced· optimizedT1 | 2024-03-07Epoch AI |
Claims drawn from cited facts, not live model generation.
This model originates from Anthropic. Released in 2024, it is a recent vintage rather than a legacy system, which suggests it reflects comparatively current training and architectural practices.
2 cited facts
This model is scored on 10 tracked benchmarks, but it currently holds no top score on any of them. Because the record has been independently reproduced, those results are backed by outside evaluation rather than resting on vendor claims alone.
3 cited facts
On average, this model trails the best reported score by 62.15 points under the disclosed evaluation harness — a wide gap that reads as well behind the leaders rather than at the frontier or in the competitive pack. It holds the top score on none of its benchmarks, so its position reflects no category leadership.
2 cited facts
6 facts cross-checked across data sources: 6 single-source.
Source facts, citations, and refresh stamp for this record.
Sources: epoch_benchmarks, litellm_prices
Claude 3 Haiku is an AI model developed by Anthropic, released 2024-03-07. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
Claude 3 Haiku has recorded scores on 10 benchmarks, each shown with its evidence status.
Claude 3 Haiku has recorded scores on 10 benchmarks — Winogrande, GPQA diamond, MATH level 5, OTIS Mock AIME 2024-2025, DTBench, WeirdML, and 4 more. The full table above shows each score with its evidence status.
3 of 10 recorded scores (30%) are independently reproduced rather than self-reported by the lab.
Claude 3 Haiku has 10 tracked claims: 3 independently reproduced, 1 self-reported, 6 unverified.
Listed API pricing: $0.25 per million input tokens, $1.25 per million output tokens. See the pricing block for the full breakdown.