Skip to content

Claude Haiku 4.5 vs GPT-5.6 Luna

Facts only. This page does not pick a winner. For a picked winner, see the use case comparison on Tools.

Claude Haiku 4.5GPT-5.6 Luna
API model IDsclaude-haiku-4-5-20251001, claude-haiku-4-5gpt-5.6-luna
Context window200,0001,050,000
Lifecycleactiveactive
Licenseproprietaryproprietary
Trains on API dataNoNo
Zero data retentionon-requeston-request
Input per 1M tokens$1 (verified 2026-09-08)$0.2 (verified 2026-09-08)

Frequently asked questions

Which has the larger context window, Claude Haiku 4.5 or GPT-5.6 Luna?
Claude Haiku 4.5 holds 200,000 tokens and GPT-5.6 Luna holds 1,050,000. GPT-5.6 Luna holds more. The context window counts the prompt, any attached files, and the reply together.
Which is cheaper, Claude Haiku 4.5 or GPT-5.6 Luna?
Per million input tokens, Claude Haiku 4.5 was $1 (verified 2026-09-08) and GPT-5.6 Luna was $0.2 (verified 2026-09-08). GPT-5.6 Luna is cheaper per input token. Output tokens cost $5 and $1.2 respectively, and output volume usually decides the bill.
Does Claude Haiku 4.5 or GPT-5.6 Luna train on API data?
Claude Haiku 4.5 does not train on API data by default, and GPT-5.6 Luna does not train on API data by default. Zero data retention is on-request for Claude Haiku 4.5 and on-request for GPT-5.6 Luna. Verified 2026-09-08 and 2026-09-08 against the providers' own documentation.
Is Claude Haiku 4.5 or GPT-5.6 Luna being retired?
Claude Haiku 4.5 is active. GPT-5.6 Luna is active. Lifecycle checked 2026-09-08 and 2026-09-08. Pin a dated snapshot ID rather than an alias if a retirement would break you.