Luna's 2026 pricing against the previous budget champion.
gpt-5.6-luna · openai
claude-haiku-4-5 · anthropic
Benchmark
| Workload Scenario | GPT-5.6 Luna | Claude Haiku 4.5 | Price Delta |
|---|---|---|---|
| 1M input tokens (raw text) | $0.20 | $1.00 | −$0.80 |
| 1M output tokens (generation) | $1.20 | $5.00 | −$3.80 |
| 1M tokens · 70% input / 30% output mix | $0.50 | $2.20 | −$1.70 |
| Standard chat turn (4K in / 800 out, 50% cached) | $0.0014 | $0.0062 | −$0.0048 |
| Monthly scale (10K requests / day) | $420.00 | $1,860.00 | −$1,440.00 |
Negative difference = GPT-5.6 Luna is cheaper. Positive = Claude Haiku 4.5 is cheaper.
Scaling Curve
Total cost of a token volume at 70/30 input/output split (uncached). Log-log scale.
Verdict
GPT-5.6 Luna is cheaper on both input and output rates, so it costs less at every input/output mix. Price alone still isn't the whole decision: capability, latency and context limits (GPT-5.6 Luna: 1.1M, Claude Haiku 4.5: 200K) may justify the premium for your task.
FAQ
On input, GPT-5.6 Luna is cheaper ($0.20/M vs $1.00/M). On output, GPT-5.6 Luna is cheaper ($1.20/M vs $5.00/M). The same model is cheaper on both sides, so it wins at every mix.
A chat-style request (4,000 input + 800 output tokens, 50% cached) costs $0.0014 on GPT-5.6 Luna and $0.0062 on Claude Haiku 4.5 — GPT-5.6 Luna is 4.4× cheaper for that workload.
GPT-5.6 Luna supports 1,050,000 tokens (128K max output); Claude Haiku 4.5 supports 200,000 (64K max output). GPT-5.6 Luna fits 5.3× more context, which matters for long documents and agents.