GPT-5.6 Luna vs. Claude Haiku 4.5: API Cost Comparison
On this API, GPT-5.6 Luna costs $0.07 / $0.42 per 1M tokens (in/out) and Claude Haiku 4.5 costs $0.5 / $2.5. GPT-5.6 Luna is the cheaper option, and both run on the same key — switching is a one-line model parameter change.
Per-token pricing
| GPT-5.6 Luna | Claude Haiku 4.5 | |
|---|---|---|
| Vendor | OpenAI | Anthropic |
| Input · per 1M tokens · Official list | $0.2 | $1 |
| Input · per 1M tokens · Your price | $0.07 | $0.5 |
| Output · per 1M tokens · Official list | $1.2 | $5 |
| Output · per 1M tokens · Your price | $0.42 | $2.5 |
| Discount vs. list | -65% | -50% |
| Context window | 400K tokens | 200K tokens |
| Max output tokens | 128K tokens | 64K tokens |
All prices in USD per 1M tokens.
What a real workload costs
Same request stream priced on both models — 10M input + 2M output tokens per month, a typical coding-agent load:
GPT-5.6 Luna
- At official list
- $4.4 per month
- On this API
- $1.54 per month
Claude Haiku 4.5
- At official list
- $20 per month
- On this API
- $10 per month
Switching between them
Both models answer on the same endpoint. Keep your base URL and key; change only the model parameter to reroute a request:
curl https://kktt.net/v1/chat/completions \
-H "Authorization: Bearer $API_KEY" \
-d '{"model": "gpt-5.6-luna", "messages": [...]}'curl https://kktt.net/v1/chat/completions \
-H "Authorization: Bearer $API_KEY" \
-d '{"model": "claude-haiku-4-5", "messages": [...]}'FAQ
Which is cheaper, GPT-5.6 Luna or Claude Haiku 4.5?
On this API, GPT-5.6 Luna is cheaper: $0.07 in / $0.42 out per 1M tokens, versus $0.5 in / $2.5 out for Claude Haiku 4.5.
Can I use GPT-5.6 Luna and Claude Haiku 4.5 with the same API key?
Yes. Both are callable with one key on the same base URL — switching between them is a one-line change of the model parameter, which makes A/B testing or fallback routing trivial.
Do GPT-5.6 Luna and Claude Haiku 4.5 differ in context window?
GPT-5.6 Luna supports 400K of context with 128K max output; Claude Haiku 4.5 supports 200K with 64K max output.