From f4568a20c7123684ea5602700a02de6dc8091029 Mon Sep 17 00:00:00 2001 From: 404oops Date: Mon, 16 Nov 2026 17:01:32 +0100 Subject: [PATCH] Use regular GPT-5.6 Sol token prices --- .../chat/OpenAiChatResponsesProvider.test.ts | 16 ++++++---------- .../drivers/ai/providers/openai/chat/models.ts | 10 +++++----- src/docs/src/AI/chat.md | 2 ++ 3 files changed, 13 insertions(+), 15 deletions(-) diff --git a/src/backend/drivers/ai/providers/openai/chat/OpenAiChatResponsesProvider.test.ts b/src/backend/drivers/ai/providers/openai/chat/OpenAiChatResponsesProvider.test.ts index 607e6d564..42071c84f 100644 --- a/src/backend/drivers/ai/providers/openai/chat/OpenAiChatResponsesProvider.test.ts +++ b/src/backend/drivers/ai/providers/openai/chat/OpenAiChatResponsesProvider.test.ts @@ -813,8 +813,6 @@ describe('OpenAiResponsesChatProvider.complete non-stream output', () => { cached_tokens: 10, }); - // gpt-5.6-sol costs: prompt=400, completion=2000, cached=40. - const sol = OPEN_AI_MODELS.find((m) => m.id === 'gpt-5.6-sol')!; expect(recordSpy).toHaveBeenCalledTimes(1); const [usage, actor, prefix, overrides] = recordSpy.mock.calls[0]!; expect(actor).toBe(SYSTEM_ACTOR); @@ -825,9 +823,9 @@ describe('OpenAiResponsesChatProvider.complete non-stream output', () => { cached_tokens: 10, }); expect(overrides).toEqual({ - prompt_tokens: 90 * Number(sol.costs.prompt_tokens), - completion_tokens: 50 * Number(sol.costs.completion_tokens), - cached_tokens: 10 * Number(sol.costs.cached_tokens ?? 0), + prompt_tokens: 90 * 500, + completion_tokens: 50 * 3000, + cached_tokens: 10 * 50, }); }); @@ -1168,15 +1166,13 @@ describe('OpenAiResponsesChatProvider.complete streaming', () => { cached_tokens: 1, }); - // gpt-5.6-sol costs: prompt=400, completion=2000, cached=40. - const sol = OPEN_AI_MODELS.find((m) => m.id === 'gpt-5.6-sol')!; expect(recordSpy).toHaveBeenCalledTimes(1); const [, , prefix, overrides] = recordSpy.mock.calls[0]!; expect(prefix).toBe('openai:gpt-5.6-sol'); expect(overrides).toEqual({ - prompt_tokens: 3 * Number(sol.costs.prompt_tokens), - completion_tokens: 2 * Number(sol.costs.completion_tokens), - cached_tokens: 1 * Number(sol.costs.cached_tokens ?? 0), + prompt_tokens: 3 * 500, + completion_tokens: 2 * 3000, + cached_tokens: 1 * 50, }); }); diff --git a/src/backend/drivers/ai/providers/openai/chat/models.ts b/src/backend/drivers/ai/providers/openai/chat/models.ts index 95c0b6fea..e85fd4621 100644 --- a/src/backend/drivers/ai/providers/openai/chat/models.ts +++ b/src/backend/drivers/ai/providers/openai/chat/models.ts @@ -183,11 +183,11 @@ export const OPEN_AI_MODELS: IChatModel[] = [ serviceSpeedRates: { fast: 2 }, costs: { tokens: 1_000_000, - prompt_tokens: 400, - cached_tokens: 40, - cache_write_tokens: 400 * 1.25, - cache_write_12h_tokens: 400 * 2.5, - completion_tokens: 2000, + prompt_tokens: 500, + cached_tokens: 50, + cache_write_tokens: 500 * 1.25, + cache_write_12h_tokens: 500 * 2.5, + completion_tokens: 3000, }, long_context_pricing: GPT_LONG_CONTEXT_PRICING, context: 1_050_000, diff --git a/src/docs/src/AI/chat.md b/src/docs/src/AI/chat.md index 3a02e9921..fbfe21e2e 100755 --- a/src/docs/src/AI/chat.md +++ b/src/docs/src/AI/chat.md @@ -385,6 +385,8 @@ The `usageDetails.serviceSpeed` field reports the served speed. Billing uses the ### OpenAI models +GPT-5.6 Sol uses regular prices of $5 per million input tokens and $30 per million output tokens. + Passing `service_tier: "priority"` selects fast service, and `service_tier: "ultrafast"` selects ultrafast service. These requests use the same prices and supported-speed selection as `serviceSpeed`. An explicit `serviceSpeed` takes precedence. Input and reasoning tokens reported by a non-streaming Responses API call are billed even when its output is empty. These charges apply whether the request fails or retries another provider.