Compare Anthropic Claude Haiku 4.5 with other AI models
See how Claude Haiku 4.5 from Anthropic compares with other AI models on price, context length and features.
- Claude Haiku 4.5 is deprecated.
Overview
| Metric | Claude Haiku 4.5 |
|---|---|
| Author | Anthropic |
| Type | Text |
| Context length | 200K tokens |
| Max output | 64K tokens |
| Reasoning | Yes |
| Input modalities | TextImage |
| Output modalities | Text |
| Released | Oct 1, 2025 |
| Knowledge cutoff | February 2025 |
| Served by | Anthropic via Oxyy |
| API | OpenAI-compatible · Anthropic Messages/v1/chat/completions |
- Requests are sent to a healthy provider automatically; if one errors, the gateway retries against another serving the same model.
Pricing
| Metric | Claude Haiku 4.5 |
|---|---|
| Input | $0.400 /1M inputStandard $1.00 |
| Output | $2.00 /1M outputStandard $5.00 |
| Cached input | $0.040 /1M cachedStandard $0.100 |
| Cache write (5 min) | $0.500 /1M cache writeStandard $1.25 |
| Cache write (1 hour) | $0.800 /1M cache write 1hStandard $2.00 |
| Blended price (3:1 input:output) | $0.800 /1M tokens |
- Includes the Launch Promo discount (60% off the standard price).
Performance
| Metric | Claude Haiku 4.5 |
|---|---|
| Latency (p50, 7 days) | 1.63 s |
| Time to first token (p50) | 1.16 s |
| Throughput (p50) | 108.2 tok/s |
| Tokens (7 days) | 10.0M |
Visualize performance
- Claude Haiku 4.5
Features
| Metric | Claude Haiku 4.5 |
|---|---|
| Tool use | Yes |
| Structured outputs | Yes |
| Prompt caching | Yes |
| Streaming | Yes |
| Web search | Yes |
| Image input | Yes |
Activity
| Metric | Claude Haiku 4.5 |
|---|---|
| Tokens (30 days) | 45.8M tokens · last 30 days |
| Input tokens | 39.3M |
| Output tokens (excluding reasoning) | 6.5M |
| Reasoning tokens | 4.4K |
- Input 39.3M
- Output (excl. reasoning) 6.5M
- Reasoning 4.4K
Suggested models to compare
How this comparison is built
Prices are what Oxyy charges today; where a discount applies, the standard price is shown beside it. Context window, output limit, modalities and features come from the Oxyy model catalog. Latency, time to first token and throughput are medians measured on Oxyy traffic over the last 7 days (throughput from streaming requests only); a model with too little traffic shows a dash. Benchmark scores appear only with their source and date. Oxyy is an independent API gateway, not affiliated with the model vendors named on this page.