What OpenAI's 20%+ Price Cut for GPT-5.6 Sol Means — August 22, 2026
· CompaniesAutomation
Flash edition of the Radar: OpenAI cuts GPT-5.6 Sol pricing by more than 20% for developers—input from $5 to $4 and output from $30 to $20 per million tokens—across API, Codex credits, and ChatGPT Work, at least until November 21. What changes on your bill and how to take advantage of this window.
Flash edition. OpenAI is lowering the price of its flagship model GPT-5.6 Sol by more than 20% for developers, and it comes with a deadline: the price is guaranteed only until November 21. A supplement to this morning's Radar.
What happened
OpenAI announced on Friday a cut of more than 20% in what developers pay for GPT-5.6 Sol: input goes from $5 to $4 per million tokens (−20%), output from $30 to $20 (−33%), and cached input from $0.50 to $0.40, for standard short-context use. The discount applies to the API, Codex credits, and eligible ChatGPT Work plans; Pro, Plus, and Business subscriptions remain unchanged, and OpenAI guarantees the rate "at least until November 21, 2026" (Source). This is the second cut in less than a month—at the end of July they lowered Terra by 20% and Luna by up to 80%—and Reuters frames it within the growing competition from Anthropic and Chinese models (Source).
Why it matters
The price war has reached the top tier for the first time. July's cuts affected medium and cheap models; touching the star model—the one the Ramp index highlighted this week as "increasingly the choice of developers," as we covered in yesterday's flash—is applying pressure right where they are already winning. The benchmark now looks like this: Sol at $4/$20 per million tokens versus $5/$25 for Claude Opus 5 and $10/$50 for Fable 5. And the nuance that rules above all else: it's not a new rate, it's a promotion with an end date. Three months is exactly one quarter of hook-in time for you to move your workflows and then decide if you stay at whatever the new price might be.
For your business
Three moves. One: if you already use Sol via API, you don't have to do anything—the new price applies automatically—but since the output is down by a third, recalculate the cost per task for workflows that generate a lot of text (reports, summaries, code agents): they are the big winners. Two: if you ruled out the flagship model because it was too expensive, this is the window to re-test it with your bank of 20 or 30 real-world cases; at $4 and $20, the quality jump might be cheaper than manually fixing the errors of a mid-tier model. Three: budget with the post-November 21 rate, not the promotional one—if an automation's margin only works with a discount, that automation has an expiration date; keep the old price noted, set a consumption alert, and decouple the cost of your agents from the current provider.
Frequently Asked Questions
Exactly how much is the price of GPT-5.6 Sol dropping?
Input goes from $5 to $4 per million tokens and output from $30 to $20; cached input goes from $0.50 to $0.40. It applies to the API, Codex credits, and ChatGPT Work; Pro, Plus, and Business subscriptions do not change.
Is the price drop permanent?
It is not guaranteed. OpenAI secures the price "at least until November 21, 2026"; they may extend it, but today it is a three-month promotion and should be budgeted as such.
Why is OpenAI lowering prices now?
It is their second cut in less than a month, following Terra and Luna in July. Reuters attributes it to competition from Anthropic and Chinese models, and this week's adoption data shows that enterprise AI spending is moving quickly between providers.