From f0d30ac4991ee9ccff4d0da8b39a01b5eebe9138 Mon Sep 17 00:00:00 2001 From: Mark Yan Date: Sat, 22 Aug 2026 15:15:43 +0000 Subject: [PATCH] docs(sdk): context_over_200k is not always a 200K threshold --- packages/sdk/src/types.ts | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-) diff --git a/packages/sdk/src/types.ts b/packages/sdk/src/types.ts index 0e81fa652ad..e05cc1e5763 100644 --- a/packages/sdk/src/types.ts +++ b/packages/sdk/src/types.ts @@ -68,7 +68,16 @@ export interface CostTier extends Cost { /** Pricing for a provider's model, including context-size tiers. */ export interface ModelCost extends Cost { - /** Legacy compatibility field: pricing applied beyond 200K context. Prefer `tiers`. */ + /** + * Legacy compatibility field. **The threshold is not always 200K.** The generator emits this + * for any single context tier starting at 200K *or above*, so it carries the 272K rate for + * `gpt-5.5`, the 256K rate for `qwen3.6-plus` and the 512K rate for `MiniMax-M3`. Applying + * it from 200K upward over-prices every request between 200K and the real threshold - for + * `MiniMax-M3` that is a 312K-token band billed at 2x. + * + * Read `tiers[].tier.size` for the threshold that actually applies. This field exists only + * for consumers that predate `tiers`. + */ context_over_200k?: Cost /** Context-size-based pricing tiers. */ tiers?: CostTier[]