Home / Claude Haiku 5.5 vs GPT-6 Luna
Claude Haiku 5.5 vs GPT-6 Luna: price
Anthropic’s and OpenAI’s low-cost models have the same list price. The bills split on long prompts, where the two companies draw the line in different places.
Checked October 13, 2026 against Anthropic’s and OpenAI’s pricing pages
| Claude Haiku 5.5 | GPT-6 Luna | |
|---|---|---|
| Input | $0.10 | $0.10 |
| Cached input | $0.01 | $0.01 |
| Cache write | $0.135 minutes; $0.20 for 1 hour | $0.13 |
| Output | $0.50 | $0.50 |
| Batch input / output | $0.05 / $0.25 | $0.05 / $0.25 |
| Long-prompt price starts | over 100,000 tokens | over 272,000 tokens |
| Long-prompt input / output | $0.50 / $2.50 | $0.20 / $0.75 |
Cost by prompt length
| Prompt length | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| 2,000 tokens | $0.45 | $0.45 |
| 50,000 tokens | $5.25 | $5.25 |
| 120,000 tokens | $61.25long-prompt price | $12.25 |
| 300,000 tokens | $151.25long-prompt price | $60.38long-prompt price |
Up to 100,000 tokens the two cost the same. Between 100,000 and 272,000, Haiku 5.5 is on its higher price and Luna is not. Above 272,000 both are on higher prices, and Haiku 5.5’s is higher.
Your monthly cost on each
Models
- Turn on JavaScript to see the costs. The price tables below work without it.
Where else the bills differ
Tokens for the same text
Per-token prices only match if both models count your text as the same number of tokens. Anthropic says Claude 4.7 and later, Haiku 5.5 included, produce about 30% more tokens for the same text than its earlier tokenizer. OpenAI’s pricing page gives no comparable figure for GPT-6 Luna. Count a sample of your own text on both before you decide, using each company’s counter or the token totals in each API response.
How the long-prompt line is counted
For Haiku 5.5, Anthropic counts all input tokens toward the 100,000, including cache reads and writes, and prices each request on its own. OpenAI’s line for GPT-6 Luna is 272,000 input tokens.
Caching and batch
Cached input costs the same on both, a tenth of the input price. Both halve prices for batch requests, and on Haiku 5.5 the caching discount stacks with batch.
Regional processing
Keeping requests in the US costs 1.1 times the normal price on Claude 4.6 and later models. OpenAI charges 10% more on regional data-residency endpoints for models released on or after 5 March 2026.
What this page leaves out
This page compares prices only. It does not judge which model writes better answers. That depends on your task, so test both on your own examples.