@oh-my-pi/pi-catalog 18.13.1 → 18.13.3

package/CHANGELOG.md

@@ -2,6 +2,29 @@
 
 ## [Unreleased]
 
+## [18.13.3] - 2026-10-28
+
+### Added
+
+- Added DeepSeek V4.1 Flash 1017 to the Factory Droid provider for accounts with Factory's 1017 flag; turning thinking off on its Baseten route disables reasoning instead of falling back to low effort ([#17234](https://github.com/can1357/oh-my-pi/pull/17234) by [@skipper019](https://github.com/skipper019))
+
+### Changed
+
+- Made Kimi Code model discovery authoritative and cached per credential, added K3.1 Pro and K3.1 Pro Highspeed to the bundled Kimi Code catalog, and preserved live thinking metadata there ([#17224](https://github.com/can1357/oh-my-pi/pull/17224) by [@trukhinyuri](https://github.com/trukhinyuri))
+- Updated Bun engine requirement to ≥1.4.4
+
+### Fixed
+
+- Codex model refresh now lists models available to an entitled account even when other saved accounts return HTTP 402 ([#17198](https://github.com/can1357/oh-my-pi/issues/17198)).
+- LiteLLM discovery now reads every `/v2/model/info` page before selecting the proxy's model roster, rather than caching only the first 50 entries; malformed or incomplete pagination falls back without caching a partial page ([#17237](https://github.com/can1357/oh-my-pi/pull/17237) by [@maykonmichel](https://github.com/maykonmichel)).
+- Fixed hosted web search on OpenAI Responses and Codex Responses selecting GPT-5.6 Luna instead of GPT-6 Luna ([#17202](https://github.com/can1357/oh-my-pi/pull/17202) by [@mateo-bolanos](https://github.com/mateo-bolanos)).
+
+## [18.13.2] - 2026-10-27
+
+### Added
+
+- Added LM Studio decision models to discovery: models LM Studio reports as `type: "decision"` are listed as `judge` models served over its `/v1/decisions` endpoint, alongside its chat models ([#17169](https://github.com/can1357/oh-my-pi/pull/17169) by [@richiestanaway](https://github.com/richiestanaway))
+
 ## [18.13.1] - 2026-10-27
 
 ### Added

package/README.md

@@ -36,7 +36,7 @@ Model- or provider-conditional policy (identity, effort ladders, wire quirks, mo
 bun add @oh-my-pi/pi-catalog
 ```
 
-Ships TypeScript source directly (no build step); requires Bun ≥ 1.3.14.
+Ships TypeScript source directly (no build step); requires Bun ≥ 1.4.4.
 
 ## Cost calculation
 

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

@@ -27,12 +27,11 @@ export interface CodexModelDiscoveryResult {
     models: ModelSpec<"openai-codex-responses">[];
     etag?: string;
     /**
-     * Set when the backend rejected the credential itself (401/403, e.g.
-     * `token_revoked`); `models` is empty. A definitive per-account denial,
-     * unlike the `null` result for transport/parse failures, so multi-account
-     * discovery can skip the account instead of aborting.
+     * A definitive per-account denial (401/403 auth, 402 payment). `models` is
+     * empty; unlike transport/parse failures, sibling accounts can still supply
+     * an authoritative catalog.
      */
-    rejectedStatus?: 401 | 403;
+    rejectedStatus?: 401 | 402 | 403;
 }
 /**
  * Fetches model metadata from Codex backend and normalizes it for pi model management.

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

@@ -377,24 +377,32 @@ export declare const KIMI_CODE_DEFAULT_MAX_TOKENS = 32000;
 export declare function kimiCodeMaxTokens(modelId: string, fallback?: number): number;
 export declare function kimiCodeMaxTokens(modelId: string, fallback: number | null): number | null;
 export declare function kimiCodeModelManagerOptions(config?: KimiCodeModelManagerConfig): ModelManagerOptions<"openai-completions">;
-/** Native LM Studio metadata keyed by model id from `/api/v0/models`. */
+/** Native LM Studio metadata keyed by model id from `/api/v0/models` and `/api/v1/models`. */
 export interface LmStudioNativeModelMetadata {
     input: ("text" | "image")[];
     contextWindow?: number;
+    /**
+     * LM Studio reports a decision model (`type: "decision"`): it answers typed
+     * questions on `/v1/decisions` and rejects chat completions.
+     */
+    decision?: true;
 }
 /** Options for LM Studio's optional native metadata probe. */
 export interface LmStudioNativeModelMetadataOptions {
     headers?: Record<string, string>;
     signal?: AbortSignal;
 }
-/** Fetches LM Studio native model metadata used to mark VLM models as image-capable. */
+/**
+ * Fetches LM Studio native model metadata used to mark VLM models as
+ * image-capable and decision models as judges.
+ */
 export declare function fetchLmStudioNativeModelMetadata(baseUrl: string, fetchImpl?: FetchImpl, options?: LmStudioNativeModelMetadataOptions): Promise<Map<string, LmStudioNativeModelMetadata> | null>;
 export interface LmStudioModelManagerConfig {
     apiKey?: string;
     baseUrl?: string;
     fetch?: FetchImpl;
 }
-export declare function lmStudioModelManagerOptions(config?: LmStudioModelManagerConfig): ModelManagerOptions<"openai-completions">;
+export declare function lmStudioModelManagerOptions(config?: LmStudioModelManagerConfig): ModelManagerOptions<Api>;
 export interface SyntheticModelManagerConfig {
     apiKey?: string;
     baseUrl?: string;

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@oh-my-pi/pi-catalog",
-	"version": "18.13.1",
+	"version": "18.13.3",
 	"description": "Model catalog for omp: bundled model database, provider discovery descriptors, model identity, classification, and equivalence",
 	"keywords": [
 		"ai",
@@ -93,15 +93,15 @@
 		"gen:proto": "bun scripts/generate-protocols.ts"
 	},
 	"dependencies": {
-		"@oh-my-pi/omptype": "18.13.1",
-		"@oh-my-pi/pi-utils": "18.13.1"
+		"@oh-my-pi/omptype": "18.13.3",
+		"@oh-my-pi/pi-utils": "18.13.3"
 	},
 	"devDependencies": {
 		"@bgotink/kdl": "0.4.0",
-		"@oh-my-pi/pi-ai": "18.13.1",
+		"@oh-my-pi/pi-ai": "18.13.3",
 		"@types/bun": "^1.3.14"
 	},
 	"engines": {
-		"bun": ">=1.3.14"
+		"bun": ">=1.4.4"
 	}
 }

package/src/compat/rules/classes/openai.kdl

@@ -84,14 +84,14 @@ class "openai" {
 	on-api "openai-responses" {
 		revision ">=5" {
 			web-search "openai"
-			web-search-model "gpt-5.6-luna"
+			web-search-model "gpt-6-luna"
 			hosted-image #true
 			image-model "gpt-image-2"
 		}
 	}
 	on-api "openai-codex-responses" {
 		revision ">=5" {
-			web-search-model "gpt-5.6-luna"
+			web-search-model "gpt-6-luna"
 			hosted-image #true
 			image-model "gpt-image-2"
 		}

package/src/compat/rules/providers/factory-droid.kdl

@@ -345,6 +345,12 @@ provider "factory-droid" {
 			cost input=0 output=0 cache-read=0 cache-write=0
 			limits context=136464 max-tokens=65536
 		}
+		model "deepseek-v4.1-flash-1017" name="DeepSeek V4.1 Flash 1017" {
+			reasoning #true
+			input "text" "image"
+			cost input=0 output=0 cache-read=0 cache-write=0
+			limits context=908928 max-tokens=131072
+		}
 		model "deepseek-v4.1-flash" name="DeepSeek V4.1 Flash" {
 			reasoning #true
 			input "text" "image"
@@ -975,6 +981,17 @@ provider "factory-droid" {
 		thinking-requires-effort #false
 		thinking-default-level "high"
 	}
+	models "deepseek-v4.1-flash-1017" {
+		routing-family "factory"
+		upstream-rotation "fireworks" "baseten" "databricks"
+		credit-rates { input 0.12; output 4; cache-read 0.024 }
+		// No bundled Fireworks row for the 1017 slug; Fireworks prices it like V4.1 Flash.
+		list-price-from "fireworks" "deepseek-v4.1-flash"
+		thinking-efforts "low" "high" "max"
+		thinking-requires-effort #false
+		thinking-default-level "high"
+		entitlement { feature-flag "deepseek_v4_1_flash_1017" }
+	}
 	models "deepseek-v4.1-flash" {
 		routing-family "factory"
 		upstream-rotation "fireworks" "baseten" "databricks"
@@ -1131,6 +1148,13 @@ provider "factory-droid" {
 			synthetic-reasoning-content-fallback " "
 		}
 	}
+	// Droid 0.250.0 configures the 1017 checkpoint's Baseten route with
+	// thinkingMode "reasoning-effort": thinking off is sent as `none`.
+	models "deepseek-v4.1-flash-1017" {
+		on-upstream "baseten" {
+			completions-reasoning-mode "effort"
+		}
+	}
 	// The replay codec follows the selected route, including EU GLM routes.
 	on-upstream "mistral" {
 		models "glm-*" "mistral-medium-3.5" "mistral-large-4" {

package/src/compat/rules/providers/kimi-code.kdl

@@ -2,6 +2,7 @@
 
 provider "kimi-code" {
 	default-model "kimi-for-coding"
+	dynamic-models-authoritative #true
 	discovery label="Kimi Code" {
 		env "KIMI_API_KEY"
 	}
@@ -39,6 +40,11 @@ provider "kimi-code" {
 		thinking-efforts "minimal" "low" "medium" "high"
 		thinking-format "zai"
 	}
+	// K2.7 Code Highspeed always thinks (`supports_thinking_type: "only"`) but
+	// exposes no adjustable effort ladder (`think_efforts` is omitted).
+	models "kimi-for-coding-highspeed" {
+		thinking-efforts
+	}
 
 	// /coding/v1/models carries no pricing. `cost-fallback` seeds the public
 	// Moonshot list price only when discovery reports $0, so a future priced

package/src/compat/rules/runtime/behavior.kdl

@@ -100,8 +100,8 @@ behavior {
 	// (flat rate) or Standard Credits. Unknown ids have no inferred entitlement.
 	quota-tiers provider="factory-droid" {
 		tier "core" "inkling" "mistral-large-4" "mistral-medium-3.5" "glm-5.3-flash" "glm-5.3" "glm-5.2" "glm-5.2-fast" \
-			"kimi-k3" "qwen3.8-max" "nemotron-3-ultra" "deepseek-v4.1-flash" "deepseek-v4-flash-0731" \
-			"deepseek-v4-pro" "minimax-m3" "minimax-m2.7"
+			"kimi-k3" "qwen3.8-max" "nemotron-3-ultra" "deepseek-v4.1-flash" "deepseek-v4.1-flash-1017" \
+			"deepseek-v4-flash-0731" "deepseek-v4-pro" "minimax-m3" "minimax-m2.7"
 		tier "standard" "claude-fable-5.1" "claude-fable-5" "claude-opus-5-5" "claude-opus-5-5-fast" \
 			"claude-opus-5" "claude-opus-5-fast" "claude-opus-4-8" "claude-opus-4-8-fast" "claude-opus-4-7" \
 			"claude-opus-4-6" "claude-opus-4-5-20251101" "claude-sonnet-5-5" "claude-sonnet-5" \
@@ -315,8 +315,8 @@ behavior {
 			exact="gemini-3-flash-preview" exact="gemini-3-pro-image-preview"
 		route "openai-completions" exact="inkling" exact="mistral-large-4" exact="mistral-medium-3.5" exact="glm-5.3-flash" \
 			exact="glm-5.3" exact="glm-5.2" exact="glm-5.2-fast" exact="kimi-k3" exact="qwen3.8-max" \
-			exact="nemotron-3-ultra" exact="deepseek-v4.1-flash" exact="deepseek-v4-flash-0731" \
-			exact="deepseek-v4-pro" exact="minimax-m3"
+			exact="nemotron-3-ultra" exact="deepseek-v4.1-flash" exact="deepseek-v4.1-flash-1017" \
+			exact="deepseek-v4-flash-0731" exact="deepseek-v4-pro" exact="minimax-m3"
 	}
 
 	// Copilot serves these models with tighter effective limits than its