GPT-5.6 Luna vs Claude Haiku — cheap-tier list prices

Two Western “speed/cost” lanes operators shortlist after the July 30, 2026 OpenAI cut: GPT-5.6 Luna at $0.20 input / $1.20 output per 1M tokens, and Claude Haiku 4.5 at about $1 / $5. On list rates Luna undercuts Haiku on both sides — then confirm quality on your own prompts.

List-price snapshot (verify on vendor consoles)

Confirm on OpenAI pricing and Anthropic pricing. Aggregator catalogs can lag after a cut — our pricing tracker applies OpenAI official list overrides for the GPT-5.6 family when needed.

Worked cost sketch (same token budget)

Illustrative only. Assume 10M input + 5M output tokens/month, no cache hits:

• Luna: 10×$0.20 + 5×$1.20 = $8.00
• Haiku 4.5: 10×$1 + 5×$5 = $35.00

Haiku is ~4× Luna on this mix before quality, latency, or token-shape differences. Chatty tool loops amplify output-rate gaps further.

When to pick which

• Pick Luna for OpenAI-shaped agents, vision-capable cheap tier, GPT-5.6 tools/Responses, and volume where list $/1M dominates.
• Pick Haiku when Claude tone/tools/evals win, or procurement wants Anthropic on the invoice.
• Alias trap: bare gpt-5.6 routes to Sol (flagship), not Luna — pin gpt-5.6-luna.

Longer write-up with a measured task: Luna vs Haiku on CloudAxis.

FAQ

Is GPT-5.6 Luna cheaper than Claude Haiku 4.5?

At late-July 2026 list rates, yes on both input and output ($0.20/$1.20 vs about $1/$5 per 1M). Always re-check vendor consoles before budgeting.

Which model ID should I use for Luna?

Use gpt-5.6-luna. Do not use the bare gpt-5.6 alias — that routes to Sol at flagship prices.

Where can I compare live catalog rows?

Open the CloudyBot AI model pricing table and filter for gpt-5.6-luna or claude-haiku, then confirm on OpenAI and Anthropic billing pages.

Related

GPT-5.6 family pricing · Luna vs DeepSeek · Cheapest LLM API · Cost calculator. Figures are list estimates — confirm on vendor billing.