ChatGPT✓ ReviewedAnnounced 22 September 2026Reviewed 23 September 2026OpenAI cut its everyday model prices in half. The catch is measuring whether cheaper is still good enough.
OpenAI released GPT-6 Sol and GPT-6 Luna on 22 September 2026, built on GPT-6 Astra's foundations at roughly half GPT-5.6's promotional API price. Both share Astra's 1,050,000-token context window; Sol replaces GPT-5.6 Sol as the everyday reasoning workhorse and Luna replaces GPT-5.6 Luna as the fastest, cheapest option. GPT-5.6 Sol, Terra and Luna remain available during the rollout, so nothing breaks today, but the cheaper pair is now the one to test against your current setup.
What changed
OpenAI released GPT-6 Sol and GPT-6 Luna on 22 September 2026, describing both as built on the advances behind GPT-6 Astra, brought into faster and more affordable models. Both carry a 1,050,000-token context window and up to 128,000 output tokens, matching Astra. GPT-6 Sol has a 20 April 2026 knowledge cutoff and is priced at $2 per million input tokens ($0.20 cached, $2.50 per million cache-write tokens) and $10 per million output tokens, roughly half GPT-5.6 Sol's promotional rate; OpenAI positions it for complex coding and agentic workflows with stronger factual reliability than GPT-5.6 Sol. GPT-6 Luna has an 18 May 2026 knowledge cutoff and is priced at $0.10 per million input tokens ($0.01 cached) and $0.50 per million output tokens, positioned for focused, high-volume tasks such as summarisation, extraction and lightweight coding. Both rolled out the same day in ChatGPT Work and Codex, on the desktop app, web, CLI and IDE extension, for Plus, Pro, Business, Enterprise and Edu users, and in the API; Free and Go users can try GPT-6 Luna in the desktop app. GPT-5.6 Sol, Terra and Luna remain available during the rollout, so an existing workflow keeps working unchanged.
Why it matters
A model that costs half as much only helps if it still does the job. OpenAI's own framing is that Sol and Luna inherit Astra's alignment gains at a fraction of Astra's price, but that comparison is against GPT-5.6, not against your actual task. Because the older Sol, Terra and Luna remain available, there is no forced migration and no reason to switch on price alone before checking quality on real work.
Bit’s takeaway
The human take
Keep GPT-5.6 as the fallback until the new pair has proven itself on your own work, not the vendor's benchmark, and re-check the real bill rather than the headline per-token price.
Turn this into a workflow
Do not just read the change. Try it safely.
Bit starts from this reviewed update and helps you build a safe way to use it on your own work: what AI can prepare, what you still check, and where the workflow should stop.
Turn this into a safe workflow with Bit