Claude 3 Haiku
by Anthropic · anthropic/claude-3-haiku · #123 cheapest of 325 paid models
Prices updated Jul 22, 2026, 2:41 AM UTC · refreshed hourly
Claude 3 Haiku is Anthropic's fastest and most compact model for near-instant responsiveness. Quick and accurate targeted performance. See the launch announcement and benchmark results [here](https://www.anthropic.com/news/claude-3-haiku) #multimodal
Compare vs:GPT-5.6 SolGPT-5.6 TerraGPT-4.1GPT-4.1 Mini
Pricing
Input
$0.25
per 1M tokens
Output
$1.25
per 1M tokens
Blended (3:1)
$0.50
3 input : 1 output
Cache read
$0.03
per 1M cached tokens
Cache write
$0.30
per 1M tokens
Cache & batch economics
Effective prices for Claude 3 Haiku with prompt caching.
| Cache hit rate | Effective blended $/1M | Example request* | vs no cache |
|---|---|---|---|
| 0% | $0.50 | $0.0268 | — |
| 50% | $0.417 | $0.0158 | −41% |
| 90% | $0.351 | $0.00695 | −74% |
*Example: 100K sticky context + 2K new input + 1K output. Batch discount is an approximate 50% off list rates for this provider’s batch API. Confirm on the vendor’s pricing page.
What a request costs
| Workload | Input tokens | Output tokens | Cost / request | Cost / 1K requests |
|---|---|---|---|---|
| Short chat message | 500 | 300 | $0.0005 | $0.5 |
| Document summary | 8,000 | 1,000 | $0.00325 | $3.25 |
| Codebase question (RAG) | 30,000 | 2,000 | $0.01 | $10.00 |
| Long-context analysis | 150,000 | 5,000 | $0.0437 | $43.75 |
Performance
Intelligence Index
3.9
AA composite quality
Coding Index
—
Math Index
—
Agentic Index
—
Output speed
0 tok/s
median
Time to first token
0s
median
Time to first answer
0s
after reasoning tokens
Artificial Analysis benchmarks
| Benchmark | Score |
|---|---|
| GPQA Diamond | 37.4% |
| Humanity's Last Exam | 3.9% |
| LiveCodeBench | 15.4% |
| SciCode | 18.6% |
| MATH-500 | 39.4% |
| AIME | 1.0% |
| IFBench | 36.1% |
| AA-LCR (Long Context Reasoning) | 21.0% |
| Terminal-Bench Hard | 0.8% |
| τ²-Bench (Telecom) | 21.1% |
Specs
Context window
200K
tokens
Max output
4K
tokens
Input modalities
text, image
Tokenizer
Claude