@bastani/pi-ai 0.9.24 → 0.9.25-alpha.1

package/CHANGELOG.md

@@ -4,6 +4,24 @@ This package is a Bastani fork of `@earendil-works/pi-ai`. Upstream history at t
 
 ## [Unreleased]
 
+## [0.9.25-alpha.1] - 2026-09-29
+
+### Added
+
+- Added GPT-6.1 Sol for OpenAI and Codex with supported reasoning efforts, prompt-cache and tool capabilities, and long-context pricing metadata.
+- Added GPT-6.1 Sol to the GitHub Copilot catalog with Copilot's Responses endpoint, supported reasoning efforts, and provider-published context and output limits. Availability remains subject to the account's model policy.
+- Added **Sign in with ChatGPT** for the OpenAI Responses API alongside API-key authentication, separate from Codex subscription login.
+- Added Jev classifiers on Vercel AI Gateway and OpenCode Zen, and provider-reported classifier usage and costs, including billed responses with invalid answers.
+- Added the lightweight `@bastani/pi-ai/models` entry point for model catalog consumers.
+- Added `Model.serviceTiers` and `getServiceTierCost()` for the Fast and Ultrafast tiers a model advertises and their published rates. The OpenAI catalog follows OpenAI's Fast and Ultrafast pricing tables. The Codex catalog follows Codex's per-model tiers: Fast for every model except GPT-5.3 Codex Spark, and Ultrafast for GPT-6 Astra.
+- The OpenAI Responses adapter can send GPT-6 Astra's `ultrafast` tier.
+
+### Changed
+
+- The Codex Responses adapter sends `service_tier` only when the model advertises it, following Codex: `flex` always passes through, `default` sends no tier, and an unadvertised tier is left out instead of failing.
+- The OpenAI Responses adapter leaves out a Fast or Ultrafast tier the model doesn't advertise. It sends other tiers as requested.
+- Fast and Ultrafast usage is priced at the served tier's published rates. A model without tier metadata keeps the previous Fast multiplier.
+
 ## [0.9.24] - 2026-09-29
 
 ### Added

package/NOTICE.md

@@ -6,7 +6,7 @@ monorepo at `packages/ai` and publishes at the same version as `@bastani/atomic`
 
 - Upstream package: [`@earendil-works/pi-ai`](https://www.npmjs.com/package/@earendil-works/pi-ai)
 - Original fork point: `v0.84.2` (`914cf1472e715297caa30db4b9535d534a9eb718`)
-- Pi AI fixes and generated image catalog synced through audited upstream `main`: `earendil-works/pi@cb7969d212836b8939001dce159fbd2ed6ad395f`
+- Applicable Pi AI fixes and generated catalogs synced through audited upstream `main`: `earendil-works/pi@1b347794e2a630e4359f2584f4eea388145d0ddf`. Atomic retains its adaptations and does not enable upstream virtual models.
 - Catalog JSON under `src/providers/data/` is generated at build time from models.dev, matching upstream. It is not committed.
 
 Original work is Copyright (c) 2025 Mario Zechner and is licensed under the MIT License.

package/README.md

@@ -1,6 +1,6 @@
 # @bastani/pi-ai
 
-Bastani-branded fork of [`@earendil-works/pi-ai`](https://www.npmjs.com/package/@earendil-works/pi-ai) from [earendil-works/pi](https://github.com/earendil-works/pi). Originally forked at **v0.84.2** (`914cf1472e715297caa30db4b9535d534a9eb718`); upstream Pi AI fixes and the unified model catalog are synced through [`a328aa89ad6e6dc5c5628ff896769532ed3d29df`](https://github.com/earendil-works/pi/commit/a328aa89ad6e6dc5c5628ff896769532ed3d29df). `@bastani/pi-ai` publishes at the same version as Atomic. `npm run build` refreshes the models.dev catalog, same as upstream.
+Bastani-branded fork of [`@earendil-works/pi-ai`](https://www.npmjs.com/package/@earendil-works/pi-ai) from [earendil-works/pi](https://github.com/earendil-works/pi). Originally forked at **v0.84.2** (`914cf1472e715297caa30db4b9535d534a9eb718`); applicable upstream Pi AI fixes and the unified model catalog are synced through [`1b347794e2a630e4359f2584f4eea388145d0ddf`](https://github.com/earendil-works/pi/commit/1b347794e2a630e4359f2584f4eea388145d0ddf). `@bastani/pi-ai` publishes at the same version as Atomic. `npm run build` refreshes the models.dev catalog, same as upstream.
 
 The public API is a drop-in replacement: install `@bastani/pi-ai` and import from `@bastani/pi-ai` instead of `@earendil-works/pi-ai`. See [NOTICE.md](NOTICE.md). This package lives in the Atomic monorepo and publishes from `.github/workflows/publish.yml`. The first npm version must be published by hand so trusted publishing can be attached.
 
@@ -63,10 +63,10 @@ Unified LLM API with provider collections, automatic auth resolution, token and
 
 ## Supported Providers
 
-- **OpenAI**
+- **OpenAI** (API key or Sign in with ChatGPT)
 - **Ant Ling**
 - **Azure OpenAI (Responses)**
-- **OpenAI Codex** (ChatGPT Plus/Pro subscription, requires OAuth, see below)
+- **OpenAI Codex (legacy)** (ChatGPT subscription, requires OAuth, see below)
 - **Radius** (API key or OAuth, with a dynamically refreshed gateway catalog)
 - **TypeSafe** (System One classifier API)
 - **DeepSeek**
@@ -261,6 +261,8 @@ models.setProvider(openrouterProvider());
 
 Provider factories import their model catalog and a lazy API wrapper. They do not import other providers. With bundler code splitting, SDK implementations (`@anthropic-ai/sdk`, `openai`, `@google/genai`, etc.) stay in lazy chunks loaded on the first request to a model of that API.
 
+For a collection without TypeBox, built-in catalogs or provider SDKs, import `createModels` and `createProvider` from `@bastani/pi-ai/models`. Import the provider factories you need separately.
+
 ### All Built-in Providers
 
 For apps that want everything (as in Quick Start):
@@ -878,6 +880,8 @@ Classifier models consume structured JSON state and answer one or more typed que
 | `typesafe` | `jev-latest` | `TYPESAFE_API_KEY` |
 | `openrouter` | `typesafe/jev-1.13`, `~typesafe/jev-latest` | `OPENROUTER_API_KEY` or OpenRouter OAuth |
 | `cloudflare-workers-ai` | `typesafe/jev` | `CLOUDFLARE_API_KEY` and `CLOUDFLARE_ACCOUNT_ID` |
+| `vercel-ai-gateway` | `typesafe-ai/jev` | `AI_GATEWAY_API_KEY` |
+| `opencode` | `jev-1.13`, `jev-1.13-free` | `OPENCODE_API_KEY` |
 
 ```typescript
 import { builtinModels } from '@bastani/pi-ai/providers/all';
@@ -913,6 +917,8 @@ console.log(result.answers);
 
 The public contract uses `bool` questions and `{ type: "bool", probability }` answers. The TypeSafe adapter translates those to and from its `noul` wire representation. Like image generation, `classify()` resolves to a result with `stopReason: "error"` instead of rejecting for provider, authentication, or response errors.
 
+System One results include `usage` when the service reports input or output token counts. Costs use the model's catalog pricing. A billed response can retain usage even when its answers are malformed; absent usage means unknown, not zero.
+
 `ClassifierOptions.temperature` divides the answer logits by the given value before they are normalized; values above 1 soften the distribution. APIs that cannot apply it, such as System One, ignore it.
 
 ### Chat models on llama.cpp
@@ -993,6 +999,8 @@ for (const block of response.content) {
 
 `xhigh` and `max` are model-specific, opt-in levels. Use `getSupportedThinkingLevels(model)` to determine whether a concrete model exposes either level; models such as GPT-5.6 can expose both.
 
+GPT-6.1 Sol supports `low`, `medium`, `high`, `xhigh` and `max` on OpenAI and Azure Responses. It does not support `off`. The Codex provider also exposes a `minimal` UI alias that sends `low` to the API.
+
 ### Provider-Specific Options (stream/complete)
 
 `models.stream()`/`complete()` accept the owning API's full option set. Use `hasApi()` to narrow a dynamically looked-up model to its API for full option typing:
@@ -1620,7 +1628,7 @@ Browser compatibility notes:
 For small bundles, import only the providers you need:
 
 ```typescript
-import { createModels } from '@bastani/pi-ai';
+import { createModels } from '@bastani/pi-ai/models';
 import { openaiProvider } from '@bastani/pi-ai/providers/openai';
 
 const models = createModels();
@@ -1693,7 +1701,8 @@ Use this when one process needs different provider settings per request, or when
 Several providers support OAuth authentication instead of static API keys:
 
 - **Anthropic** (Claude Pro/Max subscription)
-- **OpenAI Codex** (ChatGPT Plus/Pro subscription, access to GPT-5.x Codex models)
+- **OpenAI** (Sign in with ChatGPT, direct Responses API access)
+- **OpenAI Codex (legacy)** (ChatGPT subscription, Codex Responses access)
 - **GitHub Copilot** (Copilot subscription)
 - **OpenRouter** (OAuth PKCE that mints a user-controlled API key)
 
@@ -1733,6 +1742,8 @@ await models.complete(model, context);
 await models.logout('anthropic');
 ```
 
+For OpenAI Sign in with ChatGPT, pass a fourth `LoginOptions` argument with `getDeviceId: () => installationUuid`. Persist this UUID and return the same value on later logins. The login registers a user-owned client and stores its issued client ID and granted scopes with the credential. Direct ChatGPT access omits unsupported temperature, output-token limits and cache-retention fields. If its shared subscription limit is exhausted, check [ChatGPT usage](https://chatgpt.com/settings/usage).
+
 ### Vertex AI
 
 Vertex AI models support either a Google Cloud API key or Application Default Credentials (ADC). Its provider-owned API-key login flow can configure either method:
@@ -1773,7 +1784,7 @@ Built-in login and refresh flows are private provider implementations. Use provi
 
 Provider notes:
 
-**OpenAI Codex**: Requires a ChatGPT Plus or Pro subscription. Provides access to GPT-5.x Codex models with extended context windows and reasoning capabilities. The library automatically handles session-based prompt caching when `sessionId` is provided in stream options unless `cacheRetention` is `"none"`. You can set `transport` in stream options to `"sse"`, `"websocket"`, or `"auto"` for Codex Responses transport selection. When using WebSocket with a `sessionId` and cache retention enabled, connections are reused per session and expire after 5 minutes of inactivity.
+**OpenAI Codex (legacy)**: Requires a ChatGPT subscription with model access. Includes GPT-6.1 Sol and Codex models with extended context windows and reasoning capabilities. The library handles session-based prompt caching when `sessionId` is provided unless `cacheRetention` is `"none"`. Set `transport` to `"sse"`, `"websocket"`, or `"auto"` for Codex Responses transport selection. WebSocket connections with a `sessionId` and caching enabled are reused per session and expire after 5 minutes of inactivity.
 
 Call `cleanupSessionResources(sessionId)` when finished with a Codex session so its pooled WebSocket connection does not keep the process alive. Import it from `@bastani/pi-ai`.
 

package/dist/api/cloudflare-workers-ai-system-one.js

@@ -22,7 +22,7 @@ const transport = {
     label: LABEL,
     url: (model) => new URL("run", `${model.baseUrl.replace(/\/+$/u, "")}/`),
     payload: (model, request) => ({ model: model.id, input: request }),
-    answers: (body) => {
+    output: (body) => {
         if (!isRecord(body))
             throw new Error(`${LABEL} returned an unexpected response`);
         if (body.success === false)
@@ -35,7 +35,7 @@ const transport = {
         }
         if (!isRecord(run.result))
             throw new Error(`${LABEL} returned an unexpected response`);
-        return run.result.answers;
+        return run.result;
     },
 };
 /** Cloudflare Workers AI System One classification with public `bool` values mapped to wire-level `noul`. */

package/dist/api/cloudflare-workers-ai-system-one.js.map

@@ -1 +1 @@
-{"version":3,"file":"cloudflare-workers-ai-system-one.js","sourceRoot":"","sources":["../../src/api/cloudflare-workers-ai-system-one.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,iBAAiB,EAAE,QAAQ,EAA2B,MAAM,wBAAwB,CAAC;AAE9F,MAAM,KAAK,GAAG,uBAAuB,CAAC;AAEtC,SAAS,sBAAsB,CAAC,MAAe;IAC9C,IAAI,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;QAC3B,MAAM,QAAQ,GAAG,MAAM;aACrB,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,OAAO,KAAK,CAAC,OAAO,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC;aAClG,MAAM,CAAC,CAAC,OAAO,EAAqB,EAAE,CAAC,OAAO,KAAK,SAAS,CAAC,CAAC;QAChE,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC;YAAE,OAAO,GAAG,KAAK,WAAW,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;IAC1E,CAAC;IACD,OAAO,GAAG,KAAK,iBAAiB,CAAC;AAClC,CAAC;AAED;;;;;;GAMG;AACH,MAAM,SAAS,GAAuB;IACrC,GAAG,EAAE,kCAAkC;IACvC,KAAK,EAAE,KAAK;IACZ,GAAG,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,IAAI,GAAG,CAAC,KAAK,EAAE,GAAG,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC,GAAG,CAAC;IACxE,OAAO,EAAE,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE,CAAC,CAAC,EAAE,KAAK,EAAE,KAAK,CAAC,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC;IAClE,OAAO,EAAE,CAAC,IAAI,EAAE,EAAE;QACjB,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,GAAG,KAAK,kCAAkC,CAAC,CAAC;QACjF,IAAI,IAAI,CAAC,OAAO,KAAK,KAAK;YAAE,MAAM,IAAI,KAAK,CAAC,sBAAsB,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC;QACjF,MAAM,GAAG,GAAG,IAAI,CAAC,MAAM,CAAC;QACxB,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,GAAG,KAAK,kCAAkC,CAAC,CAAC;QAChF,IAAI,GAAG,CAAC,KAAK,KAAK,WAAW,EAAE,CAAC;YAC/B,MAAM,IAAI,KAAK,CAAC,GAAG,KAAK,iCAAiC,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAChF,CAAC;QACD,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,MAAM,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,GAAG,KAAK,kCAAkC,CAAC,CAAC;QACvF,OAAO,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC;IAC3B,CAAC;CACD,CAAC;AAEF,6GAA6G;AAC7G,MAAM,CAAC,MAAM,QAAQ,GAA0C,CAAC,KAAK,EAAE,OAAO,EAAE,OAAO,EAAE,EAAE,CAC1F,iBAAiB,CAAC,SAAS,EAAE,KAAK,EAAE,OAAO,EAAE,OAAO,CAAC,CAAC","sourcesContent":["import type { ClassifierFunction, ClassifierOptions } from \"../types.ts\";\nimport { classifySystemOne, isRecord, type SystemOneTransport } from \"./system-one-shared.ts\";\n\nconst LABEL = \"Cloudflare Workers AI\";\n\nfunction cloudflareErrorMessage(errors: unknown): string {\n\tif (Array.isArray(errors)) {\n\t\tconst messages = errors\n\t\t\t.map((error) => (isRecord(error) && typeof error.message === \"string\" ? error.message : undefined))\n\t\t\t.filter((message): message is string => message !== undefined);\n\t\tif (messages.length > 0) return `${LABEL} error: ${messages.join(\"; \")}`;\n\t}\n\treturn `${LABEL} request failed`;\n}\n\n/**\n * System One models on the Workers AI REST endpoint:\n * `POST /accounts/{account}/ai/run` with `{ model, input }`. The REST API\n * wraps the model output in Cloudflare's API envelope and a run record:\n * `{ success, result: { state: \"Completed\", result: { answers, usage } } }`.\n * https://developers.cloudflare.com/ai/models/typesafe/jev/\n */\nconst transport: SystemOneTransport = {\n\tapi: \"cloudflare-workers-ai-system-one\",\n\tlabel: LABEL,\n\turl: (model) => new URL(\"run\", `${model.baseUrl.replace(/\\/+$/u, \"\")}/`),\n\tpayload: (model, request) => ({ model: model.id, input: request }),\n\tanswers: (body) => {\n\t\tif (!isRecord(body)) throw new Error(`${LABEL} returned an unexpected response`);\n\t\tif (body.success === false) throw new Error(cloudflareErrorMessage(body.errors));\n\t\tconst run = body.result;\n\t\tif (!isRecord(run)) throw new Error(`${LABEL} returned an unexpected response`);\n\t\tif (run.state !== \"Completed\") {\n\t\t\tthrow new Error(`${LABEL} run did not complete (state: ${String(run.state)})`);\n\t\t}\n\t\tif (!isRecord(run.result)) throw new Error(`${LABEL} returned an unexpected response`);\n\t\treturn run.result.answers;\n\t},\n};\n\n/** Cloudflare Workers AI System One classification with public `bool` values mapped to wire-level `noul`. */\nexport const classify: ClassifierFunction<ClassifierOptions> = (model, context, options) =>\n\tclassifySystemOne(transport, model, context, options);\n"]}
+{"version":3,"file":"cloudflare-workers-ai-system-one.js","sourceRoot":"","sources":["../../src/api/cloudflare-workers-ai-system-one.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,iBAAiB,EAAE,QAAQ,EAA2B,MAAM,wBAAwB,CAAC;AAE9F,MAAM,KAAK,GAAG,uBAAuB,CAAC;AAEtC,SAAS,sBAAsB,CAAC,MAAe;IAC9C,IAAI,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;QAC3B,MAAM,QAAQ,GAAG,MAAM;aACrB,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,OAAO,KAAK,CAAC,OAAO,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC;aAClG,MAAM,CAAC,CAAC,OAAO,EAAqB,EAAE,CAAC,OAAO,KAAK,SAAS,CAAC,CAAC;QAChE,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC;YAAE,OAAO,GAAG,KAAK,WAAW,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;IAC1E,CAAC;IACD,OAAO,GAAG,KAAK,iBAAiB,CAAC;AAClC,CAAC;AAED;;;;;;GAMG;AACH,MAAM,SAAS,GAAuB;IACrC,GAAG,EAAE,kCAAkC;IACvC,KAAK,EAAE,KAAK;IACZ,GAAG,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,IAAI,GAAG,CAAC,KAAK,EAAE,GAAG,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC,GAAG,CAAC;IACxE,OAAO,EAAE,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE,CAAC,CAAC,EAAE,KAAK,EAAE,KAAK,CAAC,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC;IAClE,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE;QAChB,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,GAAG,KAAK,kCAAkC,CAAC,CAAC;QACjF,IAAI,IAAI,CAAC,OAAO,KAAK,KAAK;YAAE,MAAM,IAAI,KAAK,CAAC,sBAAsB,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC;QACjF,MAAM,GAAG,GAAG,IAAI,CAAC,MAAM,CAAC;QACxB,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,GAAG,KAAK,kCAAkC,CAAC,CAAC;QAChF,IAAI,GAAG,CAAC,KAAK,KAAK,WAAW,EAAE,CAAC;YAC/B,MAAM,IAAI,KAAK,CAAC,GAAG,KAAK,iCAAiC,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAChF,CAAC;QACD,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,MAAM,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,GAAG,KAAK,kCAAkC,CAAC,CAAC;QACvF,OAAO,GAAG,CAAC,MAAM,CAAC;IACnB,CAAC;CACD,CAAC;AAEF,6GAA6G;AAC7G,MAAM,CAAC,MAAM,QAAQ,GAA0C,CAAC,KAAK,EAAE,OAAO,EAAE,OAAO,EAAE,EAAE,CAC1F,iBAAiB,CAAC,SAAS,EAAE,KAAK,EAAE,OAAO,EAAE,OAAO,CAAC,CAAC","sourcesContent":["import type { ClassifierFunction, ClassifierOptions } from \"../types.ts\";\nimport { classifySystemOne, isRecord, type SystemOneTransport } from \"./system-one-shared.ts\";\n\nconst LABEL = \"Cloudflare Workers AI\";\n\nfunction cloudflareErrorMessage(errors: unknown): string {\n\tif (Array.isArray(errors)) {\n\t\tconst messages = errors\n\t\t\t.map((error) => (isRecord(error) && typeof error.message === \"string\" ? error.message : undefined))\n\t\t\t.filter((message): message is string => message !== undefined);\n\t\tif (messages.length > 0) return `${LABEL} error: ${messages.join(\"; \")}`;\n\t}\n\treturn `${LABEL} request failed`;\n}\n\n/**\n * System One models on the Workers AI REST endpoint:\n * `POST /accounts/{account}/ai/run` with `{ model, input }`. The REST API\n * wraps the model output in Cloudflare's API envelope and a run record:\n * `{ success, result: { state: \"Completed\", result: { answers, usage } } }`.\n * https://developers.cloudflare.com/ai/models/typesafe/jev/\n */\nconst transport: SystemOneTransport = {\n\tapi: \"cloudflare-workers-ai-system-one\",\n\tlabel: LABEL,\n\turl: (model) => new URL(\"run\", `${model.baseUrl.replace(/\\/+$/u, \"\")}/`),\n\tpayload: (model, request) => ({ model: model.id, input: request }),\n\toutput: (body) => {\n\t\tif (!isRecord(body)) throw new Error(`${LABEL} returned an unexpected response`);\n\t\tif (body.success === false) throw new Error(cloudflareErrorMessage(body.errors));\n\t\tconst run = body.result;\n\t\tif (!isRecord(run)) throw new Error(`${LABEL} returned an unexpected response`);\n\t\tif (run.state !== \"Completed\") {\n\t\t\tthrow new Error(`${LABEL} run did not complete (state: ${String(run.state)})`);\n\t\t}\n\t\tif (!isRecord(run.result)) throw new Error(`${LABEL} returned an unexpected response`);\n\t\treturn run.result;\n\t},\n};\n\n/** Cloudflare Workers AI System One classification with public `bool` values mapped to wire-level `noul`. */\nexport const classify: ClassifierFunction<ClassifierOptions> = (model, context, options) =>\n\tclassifySystemOne(transport, model, context, options);\n"]}

package/dist/api/openai-codex-responses.d.ts

@@ -1,9 +1,9 @@
-import type { ResponseCreateParamsStreaming } from "openai/resources/responses/responses.js";
 import type { SimpleStreamOptions, StreamFunction, StreamOptions } from "../types.ts";
+import { type ResponsesServiceTier } from "./openai-responses-shared.ts";
 export interface OpenAICodexResponsesOptions extends StreamOptions {
     reasoningEffort?: "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max";
     reasoningSummary?: "auto" | "concise" | "detailed" | "off" | "on" | null;
-    serviceTier?: ResponseCreateParamsStreaming["service_tier"];
+    serviceTier?: ResponsesServiceTier;
     textVerbosity?: "low" | "medium" | "high";
     toolChoice?: "auto" | "none" | "required";
 }

package/dist/api/openai-codex-responses.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"openai-codex-responses.d.ts","sourceRoot":"","sources":["../../src/api/openai-codex-responses.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAEX,6BAA6B,EAG7B,MAAM,yCAAyC,CAAC;AAIjD,OAAO,KAAK,EAMX,mBAAmB,EACnB,cAAc,EACd,aAAa,EAGb,MAAM,aAAa,CAAC;AAgErB,MAAM,WAAW,2BAA4B,SAAQ,aAAa;IACjE,eAAe,CAAC,EAAE,MAAM,GAAG,SAAS,GAAG,KAAK,GAAG,QAAQ,GAAG,MAAM,GAAG,OAAO,GAAG,KAAK,CAAC;IACnF,gBAAgB,CAAC,EAAE,MAAM,GAAG,SAAS,GAAG,UAAU,GAAG,KAAK,GAAG,IAAI,GAAG,IAAI,CAAC;IACzE,WAAW,CAAC,EAAE,6BAA6B,CAAC,cAAc,CAAC,CAAC;IAC5D,aAAa,CAAC,EAAE,KAAK,GAAG,QAAQ,GAAG,MAAM,CAAC;IAC1C,UAAU,CAAC,EAAE,MAAM,GAAG,MAAM,GAAG,UAAU,CAAC;CAC1C;AAwJD,eAAO,MAAM,MAAM,EAAE,cAAc,CAAC,wBAAwB,EAAE,2BAA2B,CA0QxF,CAAC;AAEF,eAAO,MAAM,YAAY,EAAE,cAAc,CAAC,wBAAwB,EAAE,mBAAmB,CAqBtF,CAAC;AAkYF,MAAM,WAAW,8BAA8B;IAC9C,QAAQ,EAAE,MAAM,CAAC;IACjB,kBAAkB,EAAE,MAAM,CAAC;IAC3B,iBAAiB,EAAE,MAAM,CAAC;IAC1B,qBAAqB,EAAE,MAAM,CAAC;IAC9B,iBAAiB,EAAE,MAAM,CAAC;IAC1B,mBAAmB,EAAE,MAAM,CAAC;IAC5B,aAAa,EAAE,MAAM,CAAC;IACtB,cAAc,EAAE,MAAM,CAAC;IACvB,mBAAmB,CAAC,EAAE,MAAM,CAAC;IAC7B,sBAAsB,CAAC,EAAE,MAAM,CAAC;IAChC,iBAAiB,EAAE,MAAM,CAAC;IAC1B,YAAY,EAAE,MAAM,CAAC;IACrB,uBAAuB,CAAC,EAAE,OAAO,CAAC;IAClC,kBAAkB,CAAC,EAAE,MAAM,CAAC;CAC5B;AA0BD,wBAAgB,iCAAiC,CAAC,SAAS,EAAE,MAAM,GAAG,8BAA8B,GAAG,SAAS,CAG/G;AAED,wBAAgB,mCAAmC,CAAC,SAAS,CAAC,EAAE,MAAM,GAAG,IAAI,CAQ5E;AAED,wBAAgB,iCAAiC,CAAC,SAAS,CAAC,EAAE,MAAM,GAAG,IAAI,CAc1E"}
+{"version":3,"file":"openai-codex-responses.d.ts","sourceRoot":"","sources":["../../src/api/openai-codex-responses.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,EAMX,mBAAmB,EACnB,cAAc,EACd,aAAa,EAGb,MAAM,aAAa,CAAC;AAwBrB,OAAO,EAMN,KAAK,oBAAoB,EAGzB,MAAM,8BAA8B,CAAC;AAkCtC,MAAM,WAAW,2BAA4B,SAAQ,aAAa;IACjE,eAAe,CAAC,EAAE,MAAM,GAAG,SAAS,GAAG,KAAK,GAAG,QAAQ,GAAG,MAAM,GAAG,OAAO,GAAG,KAAK,CAAC;IACnF,gBAAgB,CAAC,EAAE,MAAM,GAAG,SAAS,GAAG,UAAU,GAAG,KAAK,GAAG,IAAI,GAAG,IAAI,CAAC;IACzE,WAAW,CAAC,EAAE,oBAAoB,CAAC;IACnC,aAAa,CAAC,EAAE,KAAK,GAAG,QAAQ,GAAG,MAAM,CAAC;IAC1C,UAAU,CAAC,EAAE,MAAM,GAAG,MAAM,GAAG,UAAU,CAAC;CAC1C;AAwJD,eAAO,MAAM,MAAM,EAAE,cAAc,CAAC,wBAAwB,EAAE,2BAA2B,CA0QxF,CAAC;AAEF,eAAO,MAAM,YAAY,EAAE,cAAc,CAAC,wBAAwB,EAAE,mBAAmB,CAqBtF,CAAC;AA2WF,MAAM,WAAW,8BAA8B;IAC9C,QAAQ,EAAE,MAAM,CAAC;IACjB,kBAAkB,EAAE,MAAM,CAAC;IAC3B,iBAAiB,EAAE,MAAM,CAAC;IAC1B,qBAAqB,EAAE,MAAM,CAAC;IAC9B,iBAAiB,EAAE,MAAM,CAAC;IAC1B,mBAAmB,EAAE,MAAM,CAAC;IAC5B,aAAa,EAAE,MAAM,CAAC;IACtB,cAAc,EAAE,MAAM,CAAC;IACvB,mBAAmB,CAAC,EAAE,MAAM,CAAC;IAC7B,sBAAsB,CAAC,EAAE,MAAM,CAAC;IAChC,iBAAiB,EAAE,MAAM,CAAC;IAC1B,YAAY,EAAE,MAAM,CAAC;IACrB,uBAAuB,CAAC,EAAE,OAAO,CAAC;IAClC,kBAAkB,CAAC,EAAE,MAAM,CAAC;CAC5B;AA0BD,wBAAgB,iCAAiC,CAAC,SAAS,EAAE,MAAM,GAAG,8BAA8B,GAAG,SAAS,CAG/G;AAED,wBAAgB,mCAAmC,CAAC,SAAS,CAAC,EAAE,MAAM,GAAG,IAAI,CAQ5E;AAED,wBAAgB,iCAAiC,CAAC,SAAS,CAAC,EAAE,MAAM,GAAG,IAAI,CAc1E"}

package/dist/api/openai-codex-responses.js

@@ -13,7 +13,7 @@ import { getDeclaredTools, getInitialSystemMessage, normalizeContext, resolveTra
 import { uuidv7 } from "../utils/uuid.js";
 import { createGrammarToolInputProperties } from "./constrained-sampling.js";
 import { clampOpenAIPromptCacheKey } from "./openai-prompt-cache.js";
-import { assertPayloadPreservesFastRoute, convertResponsesMessages, convertResponsesTools, processResponsesStream, resolveRequestedServiceTier, } from "./openai-responses-shared.js";
+import { applyServiceTierPricing, assertPayloadPreservesFastRoute, convertResponsesMessages, convertResponsesTools, processResponsesStream, resolveRequestedServiceTier, codexServiceTierForRequest, } from "./openai-responses-shared.js";
 import { buildBaseOptions } from "./simple-options.js";
 // ============================================================================
 // Configuration
@@ -409,8 +409,7 @@ function buildRequestBody(model, context, options, cacheSessionId, grammarToolIn
     if (options?.temperature !== undefined) {
         body.temperature = options.temperature;
     }
-    // A fast variant carries its own tier, so a caller that only hands over the model still routes fast.
-    const requestedServiceTier = resolveRequestedServiceTier(model, options?.serviceTier);
+    const requestedServiceTier = resolveCodexRequestServiceTier(model, options?.serviceTier);
     if (requestedServiceTier !== undefined) {
         body.service_tier = requestedServiceTier;
     }
@@ -439,31 +438,12 @@ function buildRequestBody(model, context, options, cacheSessionId, grammarToolIn
     }
     return body;
 }
-function getServiceTierCostMultiplier(model, serviceTier) {
-    // Price against the model that was actually billed upstream, so a `-fast` variant of a
-    // per-model rate (gpt-5.5) is not silently charged the generic multiplier.
-    const pricedModelId = model.fastRoute?.baseModelId ?? model.id;
-    switch (serviceTier) {
-        case "flex":
-            return 0.5;
-        case "priority":
-            return pricedModelId === "gpt-5.5" ? 2.5 : 2;
-        default:
-            return 1;
-    }
-}
-function applyServiceTierPricing(usage, serviceTier, model) {
-    const multiplier = getServiceTierCostMultiplier(model, serviceTier);
-    if (multiplier === 1)
-        return;
-    usage.cost.input *= multiplier;
-    usage.cost.output *= multiplier;
-    usage.cost.cacheRead *= multiplier;
-    usage.cost.cacheWrite *= multiplier;
-    usage.cost.total = usage.cost.input + usage.cost.output + usage.cost.cacheRead + usage.cost.cacheWrite;
+function resolveCodexRequestServiceTier(model, optionsServiceTier) {
+    return codexServiceTierForRequest(model, resolveRequestedServiceTier(model, optionsServiceTier));
 }
 function resolveCodexServiceTier(responseServiceTier, requestServiceTier) {
-    if (responseServiceTier === "default" && (requestServiceTier === "flex" || requestServiceTier === "priority")) {
+    if (responseServiceTier === "default" &&
+        (requestServiceTier === "flex" || requestServiceTier === "priority" || requestServiceTier === "ultrafast")) {
         return requestServiceTier;
     }
     return responseServiceTier ?? requestServiceTier;
@@ -491,7 +471,7 @@ function resolveCodexWebSocketUrl(baseUrl) {
 async function processStream(response, output, stream, model, grammarToolInputProperties, options, streamDeadline) {
     const events = mapCodexEvents(parseSSE(response, streamDeadline.signal), output, model, options?.onProviderStreamEvent);
     await processResponsesStream(withStreamDeadline(events, streamDeadline.deadlineMs, streamDeadline.abort), output, stream, model, {
-        serviceTier: resolveRequestedServiceTier(model, options?.serviceTier),
+        serviceTier: resolveCodexRequestServiceTier(model, options?.serviceTier),
         grammarToolInputProperties,
         resolveServiceTier: resolveCodexServiceTier,
         applyServiceTierPricing: (usage, serviceTier) => applyServiceTierPricing(usage, serviceTier, model),
@@ -1217,7 +1197,7 @@ async function processWebSocketStream(url, body, headers, output, stream, model,
     try {
         socket.send(JSON.stringify({ type: "response.create", ...requestBody }));
         await processResponsesStream(startWebSocketOutputOnFirstEvent(mapCodexEvents(parseWebSocket(socket, options?.signal, idleTimeoutMs), output, model, options?.onProviderStreamEvent), onStart), output, stream, model, {
-            serviceTier: resolveRequestedServiceTier(model, options?.serviceTier),
+            serviceTier: resolveCodexRequestServiceTier(model, options?.serviceTier),
             grammarToolInputProperties,
             resolveServiceTier: resolveCodexServiceTier,
             applyServiceTierPricing: (usage, serviceTier) => applyServiceTierPricing(usage, serviceTier, model),