Very long inputs
Choose GPT-5.6 Luna
GPT-5.6 Luna reads up to 1.05M tokens per request, against 200k for Claude Haiku 3.
Claude Haiku 3 vs GPT-5.6 Luna: price, context and capabilities
Choose GPT-5.6 Luna for very long inputs and the latest release; on the other points the two are level or the data is missing.
In short
- Claude Haiku 3 has been retired; GPT-5.6 Luna is the one you can still call.
- A price comparison is not possible: one of them has no comparable published price.
- GPT-5.6 Luna reads more per request: 1.05M tokens against 200k.
- GPT-5.6 Luna is the newer model (Jul 9, 2026, against March 2024).
- Only GPT-5.6 Luna offers reasoning, prompt caching.
Which should you choose?
Each answer below comes straight from the published prices, specifications and independent benchmark results.
Latest release
Choose GPT-5.6 Luna
GPT-5.6 Luna is the newer release (Jul 9, 2026, against March 2024 for Claude Haiku 3).
Specifications side by side
| Claude Haiku 3 | GPT-5.6 Luna | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Category | Language models | Language models |
| Status | Retired | Generally available |
| Headline price | $0.50 / 1M tokens | $0.45 / 1M tokens |
| Input | $0.25 / 1M tokens | $0.20 / 1M tokens |
| Output | $1.25 / 1M tokens | $1.20 / 1M tokens |
| Cached input | No data | $0.02 / 1M tokens |
| Context window | 200k | 1.05M |
| Max output | No data | 128k |
| Released | March 2024 | Jul 9, 2026 |
| Knowledge cutoff | No data | February 2026 |
| Licence | Proprietary | Proprietary |
| Capabilities | Vision, Tool calling, Structured output, Batch API | Vision, Reasoning, Tool calling, Structured output, Prompt caching, Batch API |
| API model id | claude-3-haiku-20240307 | gpt-5.6-luna |
Benchmarks side by side
Leader in bold.
| Benchmark | Claude Haiku 3 | GPT-5.6 Luna |
|---|---|---|
| Intelligence Index | No data | 37.3 |
| SciCode | No data | 53.6% |
| GPQA Diamond | No data | 91.1% |
| Humanity's Last Exam | No data | 39.5% |
| MMMU-Pro (visual reasoning) | No data | 78.6% |
| Long-context reasoning (AA-LCR) | No data | 83.7% |
Independent evaluations by Artificial Analysis, run the same way for every model, at the reasoning setting shown. Percentages unless marked as an index.
Frequently asked questions
Should I choose Claude Haiku 3 or GPT-5.6 Luna?
Choose GPT-5.6 Luna for very long inputs and the latest release; on the other points the two are level or the data is missing.
Which is cheaper, Claude Haiku 3 or GPT-5.6 Luna?
A price comparison is not possible: one of them has no comparable published price.
Which has the larger context window, Claude Haiku 3 or GPT-5.6 Luna?
GPT-5.6 Luna reads more per request: 1.05M tokens against 200k.
What is the difference between Claude Haiku 3 and GPT-5.6 Luna?
Claude Haiku 3 is made by Anthropic and GPT-5.6 Luna by OpenAI. Claude Haiku 3 has been retired; GPT-5.6 Luna is the one you can still call. A price comparison is not possible: one of them has no comparable published price. GPT-5.6 Luna reads more per request: 1.05M tokens against 200k. GPT-5.6 Luna is the newer model (Jul 9, 2026, against March 2024). Only GPT-5.6 Luna offers reasoning, prompt caching.
Which is newer, Claude Haiku 3 or GPT-5.6 Luna?
GPT-5.6 Luna was released on Jul 9, 2026 and Claude Haiku 3 on March 2024.