Claude Opus 4.8 vs Claude Haiku 4.5
Detailed comparison of API pricing, technical specifications, and benchmark scores. Updated 2026-06-29.
Cost at a Glance
Claude Opus 4.8 is 5× more expensive than Claude Haiku 4.5 — switching could save 80% on your monthly API bill.
At 5M input + 2M output tokens per month:
Choosing Claude Haiku 4.5 could save you $60.00/mo
Price Comparison (per 1M tokens)
| Metric | Claude Opus 4.8 | Claude Haiku 4.5 |
|---|---|---|
| Input (per 1M tokens) | $5.00 | $1.00 |
| Output (per 1M tokens) | $25.00 | $5.00 |
| Cache Write (per 1M tokens) | $6.25 | $1.25 |
| Cache Read (per 1M tokens) | $0.50 | $0.10 |
| Batch Input (per 1M tokens) | $2.50 | $0.50 |
| Batch Output (per 1M tokens) | $12.50 | $2.50 |
Prices are per 1 million tokens. Actual costs depend on your monthly token volume. Use the calculator →
Specifications
| Spec | Claude Opus 4.8 | Claude Haiku 4.5 |
|---|---|---|
| Context Window | 1M | 1M |
| Max Output Tokens | 128K | 64K |
| Knowledge Cutoff | 2026-03 | 2026-03 |
| Modalities | Text, Image | Text, Image |
| Prompt Caching | Yes | Yes |
| Batch API | Yes | Yes |
| Thinking / Reasoning | Yes | No |
Benchmark Scores
| Benchmark | Claude Opus 4.8 | Claude Haiku 4.5 |
|---|---|---|
| SWE-bench Verified | 87.6 | 55.3 |
| MMLU-Pro | 86.8 | 76.8 |
| HumanEval | 95.1 | 88.2 |
Benchmark scores from official provider reports. Higher is better. Missing scores indicate the provider has not published that benchmark yet.
Use Cases
Claude Opus 4.8
Claude Haiku 4.5
About the Providers
Claude Opus 4.8 is developed by Anthropic. Claude Haiku 4.5 is developed by Anthropic. Both models are from the same provider, representing different tiers in the Anthropic model lineup. For the most up-to-date pricing, always check the official API documentation from each provider.