Claude 3.5 Haiku
AnthropicLLMs
84.0
Performance
★ 4.4
Rating
290
Reviews
Text GenerationMultimodalProprietarySmall
About
Anthropic's upgraded fast model, significantly stronger than Claude 3 Haiku with performance approaching Claude 3.5 Sonnet on many tasks.
Strengths
Major upgrade over Claude 3 Haiku with performance closer to Sonnet on coding and reasoning. Excellent speed for interactive use cases. Cost-effective for tasks that need more capability than basic models. Supports tool use and computer use.
Specifications
- Context window
- 200,000
- Parameters
- —
Pricing
- Input cost
- $0.80 / 1M
- Output cost
- $4.00 / 1M
$0.80 input / $4.00 output per 1M tokens
Speed & Latency
- 140
- tokens/sec
- 250ms
- time to first token
- 4,000 RPM
- rate limits
Available On
Anthropic APIAmazon BedrockGoogle Vertex AI
Features
function callingstreamingvisionsystem messagescomputer use
Market Sentiment
Polymarket prediction market odds for "Best AI Model" — showing how the market values each provider over time.
Loading sentiment data...
Performance Trend
Benchmark score trends over time for the top 5 benchmarks.
Loading history...
Benchmarks
Scores from various benchmark tests; higher is better.
| Test | Score | Percentile | Source |
|---|---|---|---|
| BigBench Hard | 78.0 | p89 | seed |
| GSM8K | 86.5 | p91 | seed |
| HumanEval | 80.5 | p90 | seed |
| MATH | 65.0 | p87 | seed |
