Treat the original launch context as historical
- This article was originally published on August 21, 2026. OpenAI's current API guidance now highlights GPT-6 models while continuing to document GPT-5.6 separately. Google's release notes list Gemini 3.8 Flash as generally available from September 2, after Gemini 3.7 Flash became generally available on August 13. The original model IDs remain useful only when they match your integration and provider account.
- A launch notice or current model guide does not prove that a model is free for every account, available in every region, or better for an untested application. Verify the exact endpoint, lifecycle, pricing and account limits before migration.
Check configuration before comparing output quality
- Confirm the exact provider endpoint, model ID, API surface and account limits. Then run the same redacted coding or agent tasks against the current route and candidate route, including tool calls, structured output, retry behavior and the latency users actually experience.
- Keep model-family routing explicit. A cheaper or faster variant can be useful for high-volume sub-tasks while a stronger variant handles the smaller set of decisions that need more reasoning; the split should be measured rather than inferred from the product name.
How this news should enter the product
- FreeToken will monitor both provider-owned sources, map verified IDs and lifecycle information into model records, and keep access claims separate from capability news. A user should be able to move from this article to a model comparison, a key path and a bounded test without trusting a copied launch summary.
- Future changes to price, free access, limits or model retirement need their own dated evidence update and Watchlist event.
Frequently asked questions
Does this article claim GPT-5.6 or Gemini 3.7 Flash is free?
No. Free access and account limits must be verified separately from a capability announcement.
Are these the providers' newest models today?
No. This article records August 2026 launches; current official guidance has since highlighted GPT-6 and Gemini 3.8 Flash.
Should an existing application migrate immediately?
Only a dated retirement or breaking change creates immediate urgency; otherwise run a controlled evaluation first.
What is the first configuration detail to verify?
Use the exact API model ID and provider endpoint documented for the intended interface.
AI assistance is disclosed; every factual claim was reviewed.
Originally published August 21, 2026 and refreshed October 5, 2026 against provider-owned OpenAI API guidance and Gemini API release notes. FreeToken retained the dated launch context, added a current-status note and migration checklist, and did not infer free access or universal performance rankings.