@oh-my-pi/pi-catalog 18.2.6 → 18.2.8

package/CHANGELOG.md

@@ -2,6 +2,27 @@
 
 ## [Unreleased]
 
+## [18.2.8] - 2026-09-21
+
+### Added
+
+- Expanded OpenRouter provider support with embedding, reranking, video generation, text-to-speech, and speech-to-text capabilities, including five new speech-to-text models.
+- Added speech-to-text support to the OpenAI provider.
+
+## [18.2.7] - 2026-09-21
+
+### Added
+
+- Added model-kind and grounded-search capability metadata, along with catalogs for local inference and search-engine models.
+- Added OpenRouter image-model discovery and live TypeSafe judge-model discovery.
+- Added the `buildDiscoveredModel` helper for defining custom providers.
+- Added glob-based patterns for identity overrides.
+
+### Changed
+
+- Updated input cost for TypeSafe models to 0.042
+- Improved model routing and thinking-policy handling for llama.cpp Qwen models, Bonsai lineage aliases, and custom provider names.
+
 ## [18.2.5] - 2026-09-17
 
 ### Added
@@ -167,9 +188,9 @@
 ### Fixed
 
 - Fixed OpenCode Go/Zen live model discovery (`GET /v1/models`) missing `x-opencode-session` and omp's `User-Agent`: discovery requests now attribute with the stable install id so the requests OpenCode flags as `Bun fetch` carry the required session header.
-	- Fixed GPT-6 Astra requests through GitHub Copilot failing with an unsupported endpoint error ([#10874](https://github.com/can1357/oh-my-pi/pull/10874) by [@xpcmdshell](https://github.com/xpcmdshell)).
-	- Fixed GPT-6 Astra showing as free with a 272K-token window in the OpenAI Codex catalog by applying its documented pricing; `/extended-context` enables the wire-advertised 872K-token maximum ([#10980](https://github.com/can1357/oh-my-pi/pull/10980) by [@H4vC](https://github.com/H4vC)).
-	- Made extended-context catalog rebuilds faster by resolving each model's maximum window once per process ([#11039](https://github.com/can1357/oh-my-pi/pull/11039) by [@H4vC](https://github.com/H4vC)).
+   - Fixed GPT-6 Astra requests through GitHub Copilot failing with an unsupported endpoint error ([#10874](https://github.com/can1357/oh-my-pi/pull/10874) by [@xpcmdshell](https://github.com/xpcmdshell)).
+   - Fixed GPT-6 Astra showing as free with a 272K-token window in the OpenAI Codex catalog by applying its documented pricing; `/extended-context` enables the wire-advertised 872K-token maximum ([#10980](https://github.com/can1357/oh-my-pi/pull/10980) by [@H4vC](https://github.com/H4vC)).
+   - Made extended-context catalog rebuilds faster by resolving each model's maximum window once per process ([#11039](https://github.com/can1357/oh-my-pi/pull/11039) by [@H4vC](https://github.com/H4vC)).
 
 ## [18.1.9] - 2026-09-04
 

package/README.md

@@ -4,18 +4,18 @@ Model catalog for [oh-my-pi](https://github.com/can1357/oh-my-pi): bundled model
 
 ## What's inside
 
-| Module                          | Purpose                                                                                                     |
-| ------------------------------- | ----------------------------------------------------------------------------------------------------------- |
-| `models.json` + `models`        | Bundled model database (pricing, context windows, modalities, thinking support)                             |
-| `provider-models`               | Provider catalog descriptors (`CATALOG_PROVIDERS`), per-provider model resolution rules                     |
-| `discovery`                     | Runtime model discovery for OpenAI-compatible endpoints, Gemini, Codex, Cursor, Antigravity, Ollama         |
-| `compat/rules`                  | Checked-in KDL policy tree: taxonomy (classes/families/revisions), class/provider cascade rules, runtime behavior vocabulary; compiled by `bun run gen:compat` into the committed `rules.json` |
+| Module                          | Purpose                                                                                                                                                                                                                                      |
+| ------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| `models.json` + `models`        | Bundled model database (pricing, context windows, modalities, thinking support)                                                                                                                                                              |
+| `provider-models`               | Provider catalog descriptors (`CATALOG_PROVIDERS`), per-provider model resolution rules                                                                                                                                                      |
+| `discovery`                     | Runtime model discovery for OpenAI-compatible endpoints, Gemini, Codex, Cursor, Antigravity, Ollama                                                                                                                                          |
+| `compat/rules`                  | Checked-in KDL policy tree: taxonomy (classes/families/revisions), class/provider cascade rules, runtime behavior vocabulary; compiled by `bun run gen:compat` into the committed `rules.json`                                               |
 | `compat`                        | The rule engine: `classifyModel` (taxonomy), `resolveModelPolicy` (cascade), behavior accessors (`api-routes`, `model-limits`, `exclude-models`, `pricing-peer`), collapse, and OpenAI/Anthropic wire builders that consume resolved records |
-| `identity`                      | Mechanical id utilities: reference resolution against the bundled index, dialects, selection priority, tokenizer families                                |
-| `model-thinking`                | Runtime thinking helpers (`getSupportedEfforts`, effort clamping/mapping, wire-id routing) over resolved model records                                   |
-| `model-manager` / `model-cache` | Runtime model registry with discovery refresh and on-disk caching                                           |
-| `wire`                          | Wire-level helpers: Codex, Gemini headers, GitHub Copilot                                                   |
-| `effort`                        | Reasoning-effort level definitions                                                                          |
+| `identity`                      | Mechanical id utilities: reference resolution against the bundled index, dialects, selection priority, tokenizer families                                                                                                                    |
+| `model-thinking`                | Runtime thinking helpers (`getSupportedEfforts`, effort clamping/mapping, wire-id routing) over resolved model records                                                                                                                       |
+| `model-manager` / `model-cache` | Runtime model registry with discovery refresh and on-disk caching                                                                                                                                                                            |
+| `wire`                          | Wire-level helpers: Codex, Gemini headers, GitHub Copilot                                                                                                                                                                                    |
+| `effort`                        | Reasoning-effort level definitions                                                                                                                                                                                                           |
 
 Import from subpaths (`@oh-my-pi/pi-catalog/<module>`) or the root barrel.
 
@@ -42,13 +42,13 @@ Ships TypeScript source directly (no build step); requires Bun ≥ 1.3.14.
 
 The `models` subpath (also exported from the root) provides timestamp-aware pricing helpers:
 
-| API | Result |
-| --- | --- |
-| `calculateCost(model, usage, timestamp?)` | Updates and returns `usage.cost` using `model.cost`. |
-| `calculateUsageCost(cost, usage, timestamp?)` | Updates and returns `usage.cost` using a `ModelCost`. |
-| `calculateUncachedInputCost(cost, promptInputTokens, timestamp?)` | Returns the cost of a fully uncached prompt. |
-| `getTimeBasedPricingPeriod(cost, timestamp?)` | Returns `"peak"`, `"off-peak"`, or `undefined` without a schedule. |
-| `getNextTimeBasedPricingTransition(cost, timestamp?)` | Returns the next actual peak/off-peak change strictly after the timestamp, or `undefined` if none exists. |
+| API                                                               | Result                                                                                                    |
+| ----------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------- |
+| `calculateCost(model, usage, timestamp?)`                         | Updates and returns `usage.cost` using `model.cost`.                                                      |
+| `calculateUsageCost(cost, usage, timestamp?)`                     | Updates and returns `usage.cost` using a `ModelCost`.                                                     |
+| `calculateUncachedInputCost(cost, promptInputTokens, timestamp?)` | Returns the cost of a fully uncached prompt.                                                              |
+| `getTimeBasedPricingPeriod(cost, timestamp?)`                     | Returns `"peak"`, `"off-peak"`, or `undefined` without a schedule.                                        |
+| `getNextTimeBasedPricingTransition(cost, timestamp?)`             | Returns the next actual peak/off-peak change strictly after the timestamp, or `undefined` if none exists. |
 
 Timestamps are Unix milliseconds; omitted timestamps use the current time for scheduled pricing. Flat token prices are unaffected. Pricing selects the latest applicable effective rate card, then its long-context tier, then the peak/off-peak multiplier. A transition query concerns the recurring tariff, not dated rate-card changes.
 

package/THIRD-PARTY-NOTICES.txt

@@ -335,43 +335,6 @@ This license allows the work and adaptations of it to be shared and used
 commercially, as long as it is attributed to Poppy Works. The font is bundled
 here (crates/pi-natives/src/fonts/Silver.ttf) as a CJK/Unicode bitmap fallback.
 
--------------------------------------------------------------------------------
-packages/utils/src/vendor/mermaid-ascii/NOTICE
-
-This directory contains an in-house Mermaid-diagram-to-ASCII renderer adapted
-from beautiful-mermaid (https://github.com/lukilabs/beautiful-mermaid), used
-under the MIT License.
-
-  Copyright (c) 2026 Craft Docs
-
-Only the ASCII rendering pipeline is ported (flowchart/state, sequence, class,
-ER, and xychart diagrams); the SVG renderer and its `elkjs` graph-layout
-dependency, the browser entry point, and the SVG theme/style modules were
-dropped. Terminal display width is reimplemented on `Bun.stringWidth`, and
-inline label formatting (HTML tags, markdown emphasis) is reduced to plain text
-for ASCII output. Layout and edge-routing logic is preserved faithfully so
-ASCII output matches the upstream package.
-
-MIT License
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
-
 -------------------------------------------------------------------------------
 packages/coding-agent/src/markit/NOTICE
 

package/dist/types/build.d.ts

@@ -6,7 +6,7 @@
  * Request handlers read fields — they never detect, parse ids, or allocate
  * compat per request.
  */
-import type { Api, Model, ModelSpec } from "./types.js";
+import { type Api, type Model, type ModelSpec } from "./types.js";
 /**
  * Applies reviewed catalog-data value corrections (`cost-patch`,
  * `limits-patch`, `long-context-cost`, `context-window-floor`,
@@ -16,6 +16,12 @@ import type { Api, Model, ModelSpec } from "./types.js";
  * limits and pricing still win.
  */
 export declare function applyCatalogCorrections(model: Pick<ModelSpec<Api>, "cost" | "contextWindow" | "maxTokens" | "input">, catalog: Record<string, unknown>): void;
+/**
+ * Build a discovered model using the backend's catalog-selected request API.
+ * The credential-bearing provider id remains unchanged while `providerType`
+ * persists the backend policy identity across cache and config round trips.
+ */
+export declare function buildDiscoveredModel(spec: ModelSpec<Api>, providerType: string): Model<Api>;
 /**
  * Build one model from an authored spec. Bundled models.json rows are fully
  * materialized by the generator and consumed directly (see `models.ts`), so

package/dist/types/compat/auth-ids.d.ts

@@ -1,4 +1,4 @@
 /** Every provider with an `auth/<id>.kdl` policy. */
-export type AuthProviderId = "abliteration" | "aiand" | "aimlapi" | "alibaba-coding-plan" | "alibaba-token-plan" | "amazon-bedrock" | "anthropic" | "azure" | "baseten" | "bedrock-mantle" | "cerebras" | "charm-hyper" | "cline-pass" | "cloudflare-ai-gateway" | "commandcode" | "coreweave" | "cursor" | "deepinfra" | "deepseek" | "devin" | "exa" | "firepass" | "fireworks" | "github-copilot" | "gitlab-duo" | "gitlab-duo-agent" | "gmi-cloud" | "google" | "google-antigravity" | "google-gemini-cli" | "google-vertex" | "groq" | "huggingface" | "kagi" | "kilo" | "kimi-code" | "litellm" | "llama.cpp" | "lm-studio" | "meta" | "minimax" | "minimax-code" | "minimax-code-cn" | "mistral" | "moonshot" | "muse-code" | "nanogpt" | "novita" | "nvidia" | "ollama" | "ollama-cloud" | "openai" | "openai-codex" | "openai-codex-device" | "opencode-go" | "opencode-zen" | "openrouter" | "parallel" | "perplexity" | "qianfan" | "qwen-portal" | "sakana" | "siliconflow" | "siliconflow-cn" | "stencil" | "synthetic" | "tavily" | "together" | "typesafe" | "umans" | "venice" | "vercel-ai-gateway" | "vllm" | "wafer-serverless" | "xai" | "xai-oauth" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "yolo-auto" | "zai" | "zai-coding-plan" | "zenmux" | "zhipu-coding-plan";
+export type AuthProviderId = "abliteration" | "aiand" | "aimlapi" | "alibaba-coding-plan" | "alibaba-token-plan" | "amazon-bedrock" | "anthropic" | "azure" | "baseten" | "bedrock-mantle" | "cerebras" | "charm-hyper" | "cline-pass" | "cloudflare-ai-gateway" | "commandcode" | "coreweave" | "cursor" | "deepinfra" | "deepseek" | "devin" | "exa" | "firepass" | "fireworks" | "github-copilot" | "gitlab-duo" | "gitlab-duo-agent" | "gmi-cloud" | "google" | "google-antigravity" | "google-gemini-cli" | "google-vertex" | "groq" | "huggingface" | "kagi" | "kilo" | "kimi-code" | "litellm" | "llama.cpp" | "lm-studio" | "local" | "meta" | "minimax" | "minimax-code" | "minimax-code-cn" | "mistral" | "moonshot" | "muse-code" | "nanogpt" | "novita" | "nvidia" | "ollama" | "ollama-cloud" | "openai" | "openai-codex" | "openai-codex-device" | "opencode-go" | "opencode-zen" | "openrouter" | "parallel" | "perplexity" | "qianfan" | "qwen-portal" | "sakana" | "siliconflow" | "siliconflow-cn" | "stencil" | "synthetic" | "tavily" | "together" | "typesafe" | "umans" | "venice" | "vercel-ai-gateway" | "vllm" | "wafer-serverless" | "web" | "xai" | "xai-oauth" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "yolo-auto" | "zai" | "zai-coding-plan" | "zenmux" | "zhipu-coding-plan";
 /** Providers whose policy declares a `login` flow (the `/login` roster). */
 export type LoginProviderId = "abliteration" | "aiand" | "alibaba-coding-plan" | "alibaba-token-plan" | "anthropic" | "baseten" | "cerebras" | "charm-hyper" | "cline-pass" | "cloudflare-ai-gateway" | "commandcode" | "coreweave" | "cursor" | "deepinfra" | "deepseek" | "devin" | "exa" | "firepass" | "fireworks" | "github-copilot" | "gitlab-duo" | "gitlab-duo-agent" | "gmi-cloud" | "google-antigravity" | "google-gemini-cli" | "huggingface" | "kagi" | "kilo" | "kimi-code" | "litellm" | "llama.cpp" | "lm-studio" | "meta" | "minimax-code" | "minimax-code-cn" | "moonshot" | "muse-code" | "nanogpt" | "novita" | "nvidia" | "ollama" | "ollama-cloud" | "openai-codex" | "openai-codex-device" | "opencode-go" | "opencode-zen" | "openrouter" | "parallel" | "perplexity" | "qianfan" | "qwen-portal" | "sakana" | "siliconflow" | "siliconflow-cn" | "stencil" | "synthetic" | "tavily" | "together" | "typesafe" | "umans" | "venice" | "vercel-ai-gateway" | "vllm" | "wafer-serverless" | "xai" | "xai-oauth" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "yolo-auto" | "zai" | "zai-coding-plan" | "zenmux" | "zhipu-coding-plan";

package/dist/types/compat/axes.d.ts

@@ -10,7 +10,7 @@
  * one of the axis's declared records.
  */
 import type { Effort } from "../effort.js";
-import type { ThinkingControlMode } from "../types.js";
+import { type ThinkingControlMode } from "../types.js";
 /** Value shape a directive accepts (see `rules/README.md`). */
 export type AxisShape = "scalar" | "array" | "object";
 /** Axis namespace: request-wire compat, thinking control surface, or catalog metadata. */

package/dist/types/compat/cascade.d.ts

@@ -15,8 +15,9 @@ export declare class AmbiguousOverlapError extends Error {
 export declare function globMatch(pattern: string, value: string): boolean;
 /**
  * Resolve wire, thinking, and catalog assignments for one structured target.
- * Exact model effort corrections can enable reasoning; absent family/revision
- * facts never satisfy selectors that require them. Returned axes are caller-owned.
+ * Exact model effort corrections and identity-scoped neutral-upgrade policies
+ * can enable reasoning; absent family/revision facts never satisfy selectors
+ * that require them. Returned axes are caller-owned.
  *
  * @throws AmbiguousOverlapError when equal-rank rules contest one axis.
  */

package/dist/types/compat/provider-ids.d.ts

@@ -1,2 +1,2 @@
 /** Every chat-model provider with a catalog entry (`default-model` in its provider rules). */
-export type KnownProvider = "abliteration" | "aiand" | "aimlapi" | "alibaba-coding-plan" | "alibaba-token-plan" | "amazon-bedrock" | "anthropic" | "azure" | "baseten" | "bedrock-mantle" | "cerebras" | "charm-hyper" | "cline-pass" | "cloudflare-ai-gateway" | "commandcode" | "coreweave" | "cursor" | "deepinfra" | "deepseek" | "devin" | "firepass" | "fireworks" | "github-copilot" | "gitlab-duo" | "gitlab-duo-agent" | "gmi-cloud" | "google" | "google-antigravity" | "google-gemini-cli" | "google-vertex" | "groq" | "huggingface" | "kilo" | "kimi-code" | "litellm" | "lm-studio" | "meta" | "minimax" | "minimax-code" | "minimax-code-cn" | "mistral" | "moonshot" | "muse-code" | "nanogpt" | "novita" | "nvidia" | "ollama" | "ollama-cloud" | "openai" | "openai-codex" | "opencode-go" | "opencode-zen" | "openrouter" | "qianfan" | "qwen-portal" | "sakana" | "siliconflow" | "siliconflow-cn" | "synthetic" | "together" | "umans" | "venice" | "vercel-ai-gateway" | "vllm" | "wafer-serverless" | "xai" | "xai-oauth" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "yolo-auto" | "zai" | "zenmux" | "zhipu-coding-plan";
+export type KnownProvider = "abliteration" | "aiand" | "aimlapi" | "alibaba-coding-plan" | "alibaba-token-plan" | "amazon-bedrock" | "anthropic" | "azure" | "baseten" | "bedrock-mantle" | "cerebras" | "charm-hyper" | "cline-pass" | "cloudflare-ai-gateway" | "commandcode" | "coreweave" | "cursor" | "deepinfra" | "deepseek" | "devin" | "firepass" | "fireworks" | "github-copilot" | "gitlab-duo" | "gitlab-duo-agent" | "gmi-cloud" | "google" | "google-antigravity" | "google-gemini-cli" | "google-vertex" | "groq" | "huggingface" | "kilo" | "kimi-code" | "litellm" | "lm-studio" | "local" | "meta" | "minimax" | "minimax-code" | "minimax-code-cn" | "mistral" | "moonshot" | "muse-code" | "nanogpt" | "novita" | "nvidia" | "ollama" | "ollama-cloud" | "openai" | "openai-codex" | "opencode-go" | "opencode-zen" | "openrouter" | "qianfan" | "qwen-portal" | "sakana" | "siliconflow" | "siliconflow-cn" | "synthetic" | "together" | "typesafe" | "umans" | "venice" | "vercel-ai-gateway" | "vllm" | "wafer-serverless" | "web" | "xai" | "xai-oauth" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "yolo-auto" | "zai" | "zenmux" | "zhipu-coding-plan";

package/dist/types/compat/resolve.d.ts

@@ -8,6 +8,8 @@ export interface ResolvedModelPolicy<TApi extends Api = Api> {
     /** Catalog-data axis assignments (longContext, priority, …) for generation. */
     catalog: Record<string, unknown>;
 }
+/** Resolve the request adapter assigned to a discovery backend before materialization. */
+export declare function resolveDiscoveryApi(spec: ModelSpec<Api>, providerType: string): Api;
 /**
  * Resolves the full policy surface for one model spec: structured identity,
  * complete compat record, thinking metadata, and catalog-data corrections.

package/dist/types/compat/taxonomy.d.ts

@@ -10,12 +10,12 @@
  */
 import type { Effort } from "../effort.js";
 import type { ModelIdentity } from "./types.js";
-/** Two classes (or two families) tie for one model id. */
+/** Two classes, families, or reviewed override patterns tie for one model id. */
 export declare class AmbiguousIdentityError extends Error {
     readonly model: string;
     readonly first: string;
     readonly second: string;
-    constructor(model: string, first: string, second: string, kind: "class" | "family");
+    constructor(model: string, first: string, second: string, kind: "class" | "family" | "override");
 }
 /** Options for {@link classifyModel}. */
 export interface ClassifyOptions {

package/dist/types/compat/types.d.ts

@@ -5,7 +5,7 @@
  * `behavior.ts`, `resolve.ts`) exposes to consumers.
  */
 import type { Effort } from "../effort.js";
-import type { KnownApi, ThinkingControlMode, TokenCost } from "../types.js";
+import type { Api, KindApiKind, ThinkingControlMode, TokenCost } from "../types.js";
 import type { RevisionOp } from "./revision.js";
 /** Class-membership matcher kinds, most to least specific. */
 export type MatcherKind = "exact" | "bounded" | "namespace" | "prefix" | "glob";
@@ -27,11 +27,9 @@ export interface CompiledRevisionPrefix {
     prefix: string;
     anywhere?: boolean;
 }
-/** One compiled reviewed identity correction. */
-export interface CompiledIdentityOverride {
+interface CompiledIdentityOverrideFields {
     id: string;
     provider?: string;
-    model: string;
     logical?: string;
     class?: string;
     family?: string;
@@ -43,6 +41,16 @@ export interface CompiledIdentityOverride {
     provenance: string;
     expiresAtMs?: number;
 }
+/** One compiled reviewed identity correction with exactly one bare-model selector. */
+export type CompiledIdentityOverride = CompiledIdentityOverrideFields & ({
+    /** Exact bare-model selector. */
+    model: string;
+    glob?: never;
+} | {
+    model?: never;
+    /** Anchored, case-insensitive bare-model glob. */
+    glob: string;
+});
 /** One compiled model class: matchers, families, revision rules, overrides. */
 export interface CompiledClass {
     id: string;
@@ -567,7 +575,7 @@ export interface CompiledProviderDiscovery {
 export interface CompiledSeedModel {
     id: string;
     name: string;
-    api: KnownApi;
+    api: Api;
     provider: string;
     baseUrl: string;
     reasoning: boolean;
@@ -592,8 +600,8 @@ export interface CompiledSeed {
     models: CompiledSeedModel[];
 }
 /**
- * One chat-model provider's catalog entry: the non-code half of what the
- * runtime and generator know about a provider. A `providers/<id>.kdl` file
+ * One model provider's catalog entry: the non-code half of what the runtime
+ * and generator know about a provider. A `providers/<id>.kdl` file
  * declares one by carrying `default-model`; files without it are wire-compat
  * only (custom provider ids such as `llama.cpp`).
  */
@@ -611,6 +619,8 @@ export interface CompiledProvider {
     skipCrossProviderReferenceFills?: boolean;
     /** Present only for providers enrolled in `generate-models.ts` discovery. */
     discovery?: CompiledProviderDiscovery;
+    /** Non-chat model kinds mapped to their runtime transport APIs. */
+    kindApis?: Partial<Record<KindApiKind, Api>>;
     /** Authored bundled rows, when the provider cannot be discovered at generation time. */
     seed?: CompiledSeed;
 }
@@ -673,3 +683,4 @@ export interface ResolvedAxes {
      */
     reasoning: boolean;
 }
+export {};

package/dist/types/discovery/index.d.ts

@@ -5,3 +5,4 @@ export * from "./gemini-cli.js";
 export * from "./gitlab-duo-workflow.js";
 export * from "./openai-compatible.js";
 export * from "./protobuf.js";
+export * from "./typesafe.js";

package/dist/types/discovery/openai-compatible.d.ts

@@ -59,6 +59,8 @@ export interface FetchOpenAICompatibleModelsOptions<TApi extends Api> {
     apiKey?: string;
     /** Additional request headers. */
     headers?: Record<string, string>;
+    /** Query parameters appended to the `/models` probe (e.g. OpenRouter's `output_modalities` roster filter). */
+    query?: Record<string, string>;
     /** Optional AbortSignal for request cancellation; caller owns its lifecycle. */
     signal?: AbortSignal;
     /**

package/dist/types/discovery/typesafe.d.ts

@@ -0,0 +1,26 @@
+import type { FetchImpl, ModelSpec } from "../types.js";
+/** Public TypeSafe API root shared by discovery and judgment requests. */
+export declare const TYPESAFE_DEFAULT_BASE_URL = "https://api.typesafe.ai";
+/** Wire card returned by TypeSafe's authenticated `GET /v1/models`. */
+export interface TypeSafeModelCard {
+    name: string;
+    description: string;
+    release_date: string;
+}
+/** Credentials and transport overrides for TypeSafe account model discovery. */
+export interface FetchTypeSafeModelsOptions {
+    apiKey: string;
+    baseUrl?: string;
+    fetch?: FetchImpl;
+    signal?: AbortSignal;
+}
+/** Validate and de-duplicate the official `{ models: [...] }` response. */
+export declare function parseTypeSafeModelCards(payload: unknown): TypeSafeModelCard[] | null;
+/**
+ * Fetch TypeSafe's account-visible judge models.
+ *
+ * Returns `null` for transport, HTTP, or envelope failures. A valid empty
+ * roster returns `[]`, allowing the model manager to distinguish it from an
+ * unavailable discovery endpoint.
+ */
+export declare function fetchTypeSafeModels(options: FetchTypeSafeModelsOptions): Promise<ModelSpec<"typesafe">[] | null>;

package/dist/types/model-manager.d.ts

@@ -1,4 +1,4 @@
-import type { Api, Model, ModelSpec, Provider } from "./types.js";
+import { type Api, type Model, type ModelSpec, type Provider } from "./types.js";
 /**
  * Controls when dynamic endpoint models should be fetched.
  */

package/dist/types/provider-models/openai-compat.d.ts

@@ -1,11 +1,12 @@
 import type { ModelManagerOptions, ModelsDevFallback } from "../model-manager.js";
 import { getBundledModels } from "../models.js";
-import type { Api, FetchImpl, ModelSpec, Provider } from "../types.js";
+import { type Api, type FetchImpl, type ModelSpec, type Provider } from "../types.js";
 import { ALIBABA_TOKEN_PLAN_BASE_URL } from "../wire/alibaba-token-plan.js";
 import type { ModelManagerConfig } from "./descriptor-types.js";
 export interface ModelsDevModel {
     id?: string;
     name?: string;
+    kind?: string;
     tool_call?: boolean;
     reasoning?: boolean;
     reasoning_options?: Array<{
@@ -26,6 +27,7 @@ export interface ModelsDevModel {
     };
     modalities?: {
         input?: string[];
+        output?: string[];
     };
     status?: string;
     provider?: {
@@ -167,8 +169,8 @@ export declare function applyXaiResponsesThinkingPolicy(model: ModelSpec<"openai
  * Render the xai-oauth KDL seed as the static runtime fallback consumed by
  * {@link xaiOAuthModelManagerOptions}.
  */
-export declare function buildXaiOAuthStaticSeed(baseUrl?: string): ModelSpec<"openai-responses">[];
-export declare function xaiOAuthModelManagerOptions(config?: XaiOAuthModelManagerConfig): ModelManagerOptions<"openai-responses">;
+export declare function buildXaiOAuthStaticSeed(baseUrl?: string): ModelSpec<Api>[];
+export declare function xaiOAuthModelManagerOptions(config?: XaiOAuthModelManagerConfig): ModelManagerOptions<Api>;
 export declare function isLikelyAimlApiChatModelId(id: string): boolean;
 export interface AimlApiModelManagerConfig {
     apiKey?: string;
@@ -322,7 +324,7 @@ export interface OpenRouterModelManagerConfig {
     baseUrl?: string;
     fetch?: FetchImpl;
 }
-export declare function openrouterModelManagerOptions(config?: OpenRouterModelManagerConfig): ModelManagerOptions<"openrouter">;
+export declare function openrouterModelManagerOptions(config?: OpenRouterModelManagerConfig): ModelManagerOptions<Api>;
 export interface ZenMuxModelManagerConfig {
     apiKey?: string;
     baseUrl?: string;

package/dist/types/provider-models/special.d.ts

@@ -55,6 +55,16 @@ export interface DevinModelManagerConfig {
     fetch?: DevinModelDiscoveryOptions["fetch"];
 }
 export declare function devinModelManagerOptions(config?: DevinModelManagerConfig): ModelManagerOptions<"devin-agent">;
+export declare function localModelManagerOptions(): ModelManagerOptions<"local-inference">;
+export declare function webModelManagerOptions(): ModelManagerOptions<"web-search">;
+/** Credentials and endpoint overrides for the TypeSafe catalog manager. */
+export interface TypeSafeModelManagerConfig {
+    apiKey?: string;
+    baseUrl?: string;
+    fetch?: FetchImpl;
+}
+/** Discover account-visible judge models while keeping the bundled offline seed. */
+export declare function typesafeModelManagerOptions(config?: TypeSafeModelManagerConfig): ModelManagerOptions<"typesafe">;
 export interface ZaiModelManagerConfig {
 }
 /** Creates model-manager options for Z.AI's mixed native and Anthropic transports. */

package/dist/types/types.d.ts

@@ -4,6 +4,19 @@ export type { FetchImpl } from "@oh-my-pi/pi-utils";
 export type { KnownProvider } from "./provider-models/descriptors.js";
 export type KnownApi = "openai-completions" | "openai-responses" | "openrouter" | "openai-codex-responses" | "azure-openai-responses" | "anthropic-messages" | "bedrock-converse-stream" | "google-generative-ai" | "google-gemini-cli" | "google-vertex" | "ollama-chat" | "cursor-agent" | "gitlab-duo-agent" | "devin-agent";
 export type Api = KnownApi | (string & {});
+/** Catalog kinds used to isolate role-specific runners from session chat models. */
+export declare const MODEL_KINDS: readonly ["chat", "tiny", "image", "tts", "stt", "search", "judge", "embedding", "rerank", "video"];
+/** Technical capability of a catalog model; absent model kinds mean chat. */
+export type ModelKind = (typeof MODEL_KINDS)[number];
+/** Kinds a provider maps to a runner transport through `kind-apis` in its KDL; discovery drops rows of these kinds when the provider declares no API. */
+export declare const KIND_API_KINDS: readonly ["image", "tts", "stt", "embedding", "rerank", "video"];
+export type KindApiKind = (typeof KIND_API_KINDS)[number];
+/** Grounding transport available to chat models selected by the web role. */
+export type WebSearchGrounding = "gemini" | "anthropic" | "codex" | "xai" | "openrouter";
+/** Non-chat runner protocols accepted by catalog seeds, outside the chat dispatch union. */
+export declare const RUNNER_APIS: readonly ["local-inference", "web-search", "typesafe", "openrouter-decisions", "openai-images", "openrouter-images", "xai-tts", "openai-speech", "openai-embeddings", "openrouter-rerank", "openrouter-video", "openai-transcriptions"];
+/** Resolve a model's kind while preserving chat semantics for existing catalog rows. */
+export declare function modelKind(model: Pick<Model, "kind">): ModelKind;
 /** Canonical thinking transport used by a model. */
 export type ThinkingControlMode = "effort" | "budget" | "google-level" | "anthropic-adaptive" | "anthropic-budget-effort";
 /** Per-model thinking capabilities used to clamp and map user-facing effort levels. */
@@ -942,6 +955,10 @@ export interface ModelCost extends TokenCost {
 export type ModelTokenizer = "claude-v3" | "claude-v47" | "claude-v5" | "claude-v5-sonnet" | "qwen3" | "deepseek-v3" | "kimi-k2" | "glm5";
 export interface Model<TApi extends Api = Api> {
     id: string;
+    /** Role-specific runner capability; omitted for ordinary chat models. */
+    kind?: ModelKind;
+    /** Grounding transport supported by this chat model. */
+    webSearch?: WebSearchGrounding;
     /**
      * Structured model identity resolved by the compat engine: vendor lineage
      * class, product family, and revision. Baked into models.json rows and
@@ -984,6 +1001,12 @@ export interface Model<TApi extends Api = Api> {
     name: string;
     api: TApi;
     provider: Provider;
+    /**
+     * Discovery backend whose catalog policy applies when it differs from the
+     * credential-bearing provider id. Persisted so cached and rebuilt custom
+     * providers retain their transport backend's policy.
+     */
+    providerType?: string;
     baseUrl: string;
     reasoning: boolean;
     /**

package/package.json

@@ -1,51 +1,25 @@
 {
-	"type": "module",
 	"name": "@oh-my-pi/pi-catalog",
-	"version": "18.2.6",
+	"version": "18.2.8",
 	"description": "Model catalog for omp: bundled model database, provider discovery descriptors, model identity, classification, and equivalence",
+	"keywords": [
+		"ai",
+		"catalog",
+		"discovery",
+		"llm",
+		"models"
+	],
 	"homepage": "https://omp.sh",
-	"author": "Stencil Labs, Inc.",
+	"bugs": {
+		"url": "https://github.com/can1357/oh-my-pi/issues"
+	},
 	"license": "MIT",
+	"author": "Stencil Labs, Inc.",
 	"repository": {
 		"type": "git",
 		"url": "git+https://github.com/can1357/oh-my-pi.git",
 		"directory": "packages/catalog"
 	},
-	"bugs": {
-		"url": "https://github.com/can1357/oh-my-pi/issues"
-	},
-	"keywords": [
-		"ai",
-		"llm",
-		"models",
-		"catalog",
-		"discovery"
-	],
-	"main": "./src/index.ts",
-	"types": "./dist/types/index.d.ts",
-	"scripts": {
-		"check": "oxlint . && oxfmt --check --no-error-on-unmatched-pattern 'src/**/*.{ts,tsx}' '{test,bench,examples,scripts}/**/*.ts' '*.ts' && bun run check:types",
-		"check:types": "tsgo -p tsconfig.json --noEmit",
-		"lint": "oxlint .",
-		"test": "bun test --parallel",
-		"fix": "oxlint --fix --fix-suggestions . && bun run fmt",
-		"fmt": "oxfmt --no-error-on-unmatched-pattern 'src/**/*.{ts,tsx}' '{test,bench,examples,scripts}/**/*.ts' '*.ts'",
-		"gen:models": "bun scripts/generate-models.ts",
-		"gen:compat": "bun scripts/compile-compat.ts",
-		"gen:proto": "bun scripts/generate-protocols.ts"
-	},
-	"dependencies": {
-		"@oh-my-pi/omptype": "18.2.6",
-		"@oh-my-pi/pi-utils": "18.2.6"
-	},
-	"devDependencies": {
-		"@bgotink/kdl": "0.4.0",
-		"@oh-my-pi/pi-ai": "18.2.6",
-		"@types/bun": "^1.3.14"
-	},
-	"engines": {
-		"bun": ">=1.3.14"
-	},
 	"files": [
 		"src",
 		"README.md",
@@ -54,6 +28,9 @@
 		"THIRD-PARTY-NOTICES.txt",
 		"dist/types"
 	],
+	"type": "module",
+	"main": "./src/index.ts",
+	"types": "./dist/types/index.d.ts",
 	"exports": {
 		".": {
 			"types": "./dist/types/index.d.ts",
@@ -100,5 +77,28 @@
 			"import": "./src/*.ts"
 		},
 		"./*.js": "./src/*.ts"
+	},
+	"scripts": {
+		"check": "oxlint . && oxfmt --check --no-error-on-unmatched-pattern 'src/**/*.{ts,tsx}' '{test,bench,examples,scripts}/**/*.ts' '*.ts' && bun run check:types",
+		"check:types": "tsgo -p tsconfig.json --noEmit",
+		"lint": "oxlint .",
+		"test": "bun test --parallel",
+		"fix": "oxlint --fix --fix-suggestions . && bun run fmt",
+		"fmt": "oxfmt --no-error-on-unmatched-pattern 'src/**/*.{ts,tsx}' '{test,bench,examples,scripts}/**/*.ts' '*.ts'",
+		"gen:models": "bun scripts/generate-models.ts",
+		"gen:compat": "bun scripts/compile-compat.ts",
+		"gen:proto": "bun scripts/generate-protocols.ts"
+	},
+	"dependencies": {
+		"@oh-my-pi/omptype": "18.2.8",
+		"@oh-my-pi/pi-utils": "18.2.8"
+	},
+	"devDependencies": {
+		"@bgotink/kdl": "0.4.0",
+		"@oh-my-pi/pi-ai": "18.2.8",
+		"@types/bun": "^1.3.14"
+	},
+	"engines": {
+		"bun": ">=1.3.14"
 	}
 }