@oh-my-pi/pi-catalog 18.11.1 → 18.11.3

package/CHANGELOG.md

@@ -2,6 +2,33 @@
 
 ## [Unreleased]
 
+## [18.11.3] - 2026-10-16
+
+### Changed
+
+- Renamed the built-in `google-antigravity` login option to "Google Antigravity (Gemini, Claude, GPT-OSS)" so it is distinguishable from similarly named third-party plugins and no longer implies support stops at Gemini 3 ([#15998](https://github.com/can1357/oh-my-pi/pull/15998) by [@savanne-kham](https://github.com/savanne-kham)).
+
+### Fixed
+
+- Fixed Command Code's `z-ai/glm-5.3-flash` failing every request with "not available on this endpoint": the catalog routes it to chat completions, the only endpoint Command Code lists for it ([#15922](https://github.com/can1357/oh-my-pi/pull/15922) by [@jasonharrison](https://github.com/jasonharrison))
+- Fixed `/handoff` failing with `400 Provider returned error` on OpenRouter's hosted Muse Spark 1.x and Muse Wonder models: requests no longer send a `tool_choice` the upstream Meta backend rejects ([#15910](https://github.com/can1357/oh-my-pi/issues/15910))
+- Fixed OpenCode Zen bringing back retired bundled models from an authoritative model cache, and dropping drop-listed models such as `gemini-3.7-flash` when a refresh fails after an upgrade ([#15770](https://github.com/can1357/oh-my-pi/pull/15770) by [@adi-IL](https://github.com/adi-IL))
+
+## [18.11.2] - 2026-10-16
+
+### Added
+
+- Added `keep-bundled-names` catalog attribute to preserve static model display names during discovery
+- Updated `openai-codex` provider configuration to prevent model name regressions during live discovery
+
+### Changed
+
+- GPT Image models (`gpt-image-*`, `chatgpt-image-*`) now classify as their own `gpt-image` family instead of `gpt` or `chatgpt`, so GPT chat-model rules no longer apply to them ([#14619](https://github.com/can1357/oh-my-pi/pull/14619) by [@jimhester](https://github.com/jimhester))
+
+### Fixed
+
+- Fixed OpenCode Zen Gemini 4 Argon failing with an unsupported protocol; it now uses the native Gemini endpoint ([#15977](https://github.com/can1357/oh-my-pi/issues/15977)).
+
 ## [18.11.1] - 2026-10-15
 
 ### Fixed

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

@@ -16,7 +16,7 @@ export type AxisShape = "scalar" | "array" | "object";
 /** Axis namespace: request-wire compat, thinking control surface, or catalog metadata. */
 export type AxisSet = "wire" | "thinking" | "catalog";
 /** Resolved compat record families a wire axis may be assigned onto. */
-export type CompatRecordName = "openai" | "openai-responses" | "anthropic" | "bedrock" | "devin" | "google" | "request";
+export type CompatRecordName = "openai" | "openai-responses" | "anthropic" | "bedrock" | "devin" | "google" | "request" | "ollama";
 /** One axis definition: resolved key, namespace, shape, and applicability. */
 export interface AxisDef {
     /** Resolved camelCase field the directive assigns. */

package/dist/types/compat/request-size.d.ts

@@ -7,7 +7,8 @@ import type { Model } from "../types.js";
  */
 export declare function resolveInlineImageByteBudget(model: Model, officialEndpoint: boolean): number | undefined;
 /**
- * Most images one request to this deployment may carry, or `undefined` when no rule names the host. The
+ * Most images one request to this deployment may carry, or `undefined` when no rule names the host. A configured
+ * `model.imageCountBudget` wins: the user knows the host behind a custom gateway. Otherwise the
  * `official-image-count-budget` axis replaces `image-count-budget` only when `officialEndpoint` says the request
  * reaches the model's official endpoint (see {@link resolveInlineImageByteBudget}); a rerouted endpoint keeps the
  * provider's generic budget.

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

@@ -634,6 +634,8 @@ export interface CompiledProvider {
     dynamicModelsAuthoritative?: boolean;
     /** Generator backfills never copy reasoning/input/limits from same-id rows on other hosts. */
     skipCrossProviderReferenceFills?: boolean;
+    /** Runtime discovery keeps a bundled row's display name instead of the live one. */
+    keepBundledNames?: boolean;
     /** Present only for providers enrolled in `generate-models.ts` discovery. */
     discovery?: CompiledProviderDiscovery;
     /** Non-chat model kinds mapped to their runtime transport APIs. */

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

@@ -51,6 +51,12 @@ export interface ModelManagerOptions<TApi extends Api = Api, TModelsDevPayload =
      * a zero discovery price is treated as unknown and the bundled rate is kept.
      */
     dynamicCostAuthoritative?: boolean;
+    /**
+     * When true, a static row's display name wins over the cached or live row
+     * for the same id, so a model keeps one name before and after discovery.
+     * Set from the provider's `keep-bundled-names` catalog entry.
+     */
+    keepStaticNames?: boolean;
     /**
      * When true, a fresh cache never satisfies an online-eligible refresh: the
      * dynamic fetch always runs (an explicit `"offline"` strategy is still

package/dist/types/types.d.ts

@@ -1100,10 +1100,16 @@ export type ResolvedGoogleCompat = Required<Omit<GoogleCompat, "streamFirstEvent
     maxImageDimension?: number;
     maxImagePayloadBytes?: number;
 };
+/** Ollama chat history compatibility resolved by the catalog. */
+export interface OllamaCompat {
+    /** Omit assistant thinking when the deployment rejects it in history. */
+    stripThinkingHistory?: boolean;
+}
+export type ResolvedOllamaCompat = Required<OllamaCompat>;
 /** Sparse, user-authored compat overrides for a given API (models.json / config vocabulary). */
-export type CompatConfigOf<TApi extends Api> = TApi extends "openai-completions" | "openrouter" | "openai-responses" | "azure-openai-responses" | "openai-codex-responses" ? OpenAICompat : TApi extends "anthropic-messages" ? AnthropicCompat : TApi extends "bedrock-converse-stream" ? BedrockCompat : TApi extends "devin-agent" ? DevinCompat : TApi extends "google-generative-ai" | "google-vertex" | "google-gemini-cli" ? GoogleCompat : undefined;
+export type CompatConfigOf<TApi extends Api> = TApi extends "openai-completions" | "openrouter" | "openai-responses" | "azure-openai-responses" | "openai-codex-responses" ? OpenAICompat : TApi extends "anthropic-messages" ? AnthropicCompat : TApi extends "bedrock-converse-stream" ? BedrockCompat : TApi extends "devin-agent" ? DevinCompat : TApi extends "google-generative-ai" | "google-vertex" | "google-gemini-cli" ? GoogleCompat : TApi extends "ollama-chat" ? OllamaCompat : undefined;
 /** Resolved compat for a given API: complete record, materialized once by `buildModel`. */
-export type CompatOf<TApi extends Api> = TApi extends "openrouter" ? ResolvedOpenRouterCompat : TApi extends "openai-completions" ? ResolvedOpenAICompat : TApi extends "openai-responses" | "azure-openai-responses" | "openai-codex-responses" ? ResolvedOpenAIResponsesCompat : TApi extends "anthropic-messages" ? ResolvedAnthropicCompat : TApi extends "bedrock-converse-stream" ? ResolvedBedrockCompat : TApi extends "devin-agent" ? ResolvedDevinCompat : TApi extends "google-generative-ai" | "google-vertex" | "google-gemini-cli" ? ResolvedGoogleCompat : undefined;
+export type CompatOf<TApi extends Api> = TApi extends "openrouter" ? ResolvedOpenRouterCompat : TApi extends "openai-completions" ? ResolvedOpenAICompat : TApi extends "openai-responses" | "azure-openai-responses" | "openai-codex-responses" ? ResolvedOpenAIResponsesCompat : TApi extends "anthropic-messages" ? ResolvedAnthropicCompat : TApi extends "bedrock-converse-stream" ? ResolvedBedrockCompat : TApi extends "devin-agent" ? ResolvedDevinCompat : TApi extends "google-generative-ai" | "google-vertex" | "google-gemini-cli" ? ResolvedGoogleCompat : TApi extends "ollama-chat" ? ResolvedOllamaCompat : undefined;
 /** Provider-native compaction endpoint configuration for one model. */
 export interface RemoteCompactionConfig<TApi extends Api = Api> {
     /** Enables provider-native compaction for providers not enabled by built-in policy. */
@@ -1280,6 +1286,12 @@ export interface Model<TApi extends Api = Api> {
      * than OMP's general image pipeline. `stb` local backends reject WebP.
      */
     imageInputDecoder?: "stb";
+    /**
+     * Most images one request may carry, from user configuration. Replaces the
+     * catalog's `image-count-budget` for hosts no rule names, such as a custom
+     * gateway in front of a vendor with a higher limit.
+     */
+    imageCountBudget?: number;
     /**
      * Native provider tool-call support. `false` is the only unsupported signal:
      * `true` and `undefined` both mean callers may use native tools. Catalog and

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@oh-my-pi/pi-catalog",
-	"version": "18.11.1",
+	"version": "18.11.3",
 	"description": "Model catalog for omp: bundled model database, provider discovery descriptors, model identity, classification, and equivalence",
 	"keywords": [
 		"ai",
@@ -93,12 +93,12 @@
 		"gen:proto": "bun scripts/generate-protocols.ts"
 	},
 	"dependencies": {
-		"@oh-my-pi/omptype": "18.11.1",
-		"@oh-my-pi/pi-utils": "18.11.1"
+		"@oh-my-pi/omptype": "18.11.3",
+		"@oh-my-pi/pi-utils": "18.11.3"
 	},
 	"devDependencies": {
 		"@bgotink/kdl": "0.4.0",
-		"@oh-my-pi/pi-ai": "18.11.1",
+		"@oh-my-pi/pi-ai": "18.11.3",
 		"@types/bun": "^1.3.14"
 	},
 	"engines": {

package/src/compat/axes.ts

@@ -20,7 +20,15 @@ export type AxisShape = "scalar" | "array" | "object";
 export type AxisSet = "wire" | "thinking" | "catalog";
 
 /** Resolved compat record families a wire axis may be assigned onto. */
-export type CompatRecordName = "openai" | "openai-responses" | "anthropic" | "bedrock" | "devin" | "google" | "request";
+export type CompatRecordName =
+	| "openai"
+	| "openai-responses"
+	| "anthropic"
+	| "bedrock"
+	| "devin"
+	| "google"
+	| "request"
+	| "ollama";
 
 /** One axis definition: resolved key, namespace, shape, and applicability. */
 export interface AxisDef {
@@ -251,7 +259,7 @@ export const AXES: Readonly<Record<string, AxisDef>> = {
 	"requires-thinking-enabled": wire("requiresThinkingEnabled", ["anthropic"]),
 	"requires-tool-result-id": wire("requiresToolResultId", ["anthropic"]),
 	"signing-endpoint": wire("signingEndpoint", ["anthropic"]),
-	"strip-thinking-history": wire("stripThinkingHistory", ["anthropic"]),
+	"strip-thinking-history": wire("stripThinkingHistory", ["anthropic", "ollama"]),
 	"supports-context-management": wire("supportsContextManagement", ["anthropic"]),
 	"supports-output-effort": wire("supportsOutputEffort", ["anthropic"]),
 	"supports-eager-tool-input-streaming": wire("supportsEagerToolInputStreaming", ["anthropic"]),
@@ -596,6 +604,7 @@ export const AXES: Readonly<Record<string, AxisDef>> = {
 
 /** Records applicable to each API family; used by `resolve.ts` when applying wire axes. */
 export const API_COMPAT_RECORDS: Readonly<Record<string, readonly CompatRecordName[]>> = {
+	"ollama-chat": ["ollama"],
 	"openai-completions": ["openai"],
 	openrouter: ["openai", "openai-responses"],
 	"openai-responses": ["openai-responses"],

package/src/compat/request-size.ts

@@ -26,12 +26,14 @@ export function resolveInlineImageByteBudget(model: Model, officialEndpoint: boo
 }
 
 /**
- * Most images one request to this deployment may carry, or `undefined` when no rule names the host. The
+ * Most images one request to this deployment may carry, or `undefined` when no rule names the host. A configured
+ * `model.imageCountBudget` wins: the user knows the host behind a custom gateway. Otherwise the
  * `official-image-count-budget` axis replaces `image-count-budget` only when `officialEndpoint` says the request
  * reaches the model's official endpoint (see {@link resolveInlineImageByteBudget}); a rerouted endpoint keeps the
  * provider's generic budget.
  */
 export function resolveImageCountBudget(model: Model, officialEndpoint: boolean): number | undefined {
+	if (model.imageCountBudget !== undefined) return model.imageCountBudget;
 	const policy = resolveCatalogPolicy(model);
 	const value = (officialEndpoint && policy.officialImageCountBudget) || policy.imageCountBudget;
 	return typeof value === "number" && Number.isInteger(value) && value > 0 ? value : undefined;

package/src/compat/resolve.ts

@@ -27,6 +27,7 @@ import type {
 	ResolvedBedrockCompat,
 	ResolvedDevinCompat,
 	ResolvedGoogleCompat,
+	ResolvedOllamaCompat,
 	ResolvedOpenAICompat,
 	ResolvedOpenAIResponsesCompat,
 	ResolvedOpenAISharedCompat,
@@ -974,6 +975,13 @@ function resolveBedrockPolicy(spec: ModelSpec<"bedrock-converse-stream">, axes:
 	return compat;
 }
 
+function resolveOllamaPolicy(spec: ModelSpec<"ollama-chat">, axes: ResolvedAxes): ResolvedOllamaCompat {
+	const compat: ResolvedOllamaCompat = { stripThinkingHistory: false };
+	applyWireAxes(compat, axes.wire, "ollama-chat");
+	applyCompatOverrides(compat, spec.compat);
+	return compat;
+}
+
 function resolveDevinPolicy(spec: ModelSpec<"devin-agent">, axes: ResolvedAxes): ResolvedDevinCompat {
 	const compat: ResolvedDevinCompat = {
 		trustExplicitThinkingOnly: true,
@@ -1380,6 +1388,8 @@ export function resolveModelPolicy(input: ModelSpec<Api>, route?: ResolveRoute):
 		compat = resolveAnthropicPolicy(spec, facts, axes);
 	} else if (specUsesApi(spec, "bedrock-converse-stream")) {
 		compat = resolveBedrockPolicy(spec, axes);
+	} else if (specUsesApi(spec, "ollama-chat")) {
+		compat = resolveOllamaPolicy(spec, axes);
 	} else if (specUsesApi(spec, "devin-agent")) {
 		compat = resolveDevinPolicy(spec, axes);
 	} else if (

package/src/compat/rules/README.md

@@ -403,6 +403,7 @@ provider "sakana" {
     dynamic-models-authoritative #true            // discovery replaces bundled rows
     allow-unauthenticated #true                   // runtime manager without a key
     skip-cross-provider-reference-fills #true     // generator never backfills from same-id rows elsewhere
+    keep-bundled-names #true                      // discovery never renames a bundled row
     discovery label="Sakana AI" oauth-provider="sakana" allow-unauthenticated=#true {
         env "SAKANA_GEN_KEY"                      // generation-time keys; defaults to the provider env
     }

package/src/compat/rules/auth/google-antigravity.kdl

@@ -1,5 +1,5 @@
 auth "google-antigravity" {
-	name "Antigravity (Gemini 3, Claude, GPT-OSS)"
+	name "Google Antigravity (Gemini, Claude, GPT-OSS)"
 	login "oauth-code" {
 		// Antigravity uses different Google OAuth credentials for its additional models.
 		client-id "MTA3MTAwNjA2MDU5MS10bWhzc2luMmgyMWxjcmUyMzV2dG9sb2poNGc0MDNlcC5hcHBzLmdvb2dsZXVzZXJjb250ZW50LmNvbQ==" encoding="base64"

package/src/compat/rules/providers/ollama-cloud.kdl

@@ -5,6 +5,8 @@ provider "ollama-cloud" {
 	env "OLLAMA_CLOUD_API_KEY"
 	discovery label="Ollama Cloud" oauth-provider="ollama-cloud"
 
+	// Legacy deployments can reject assistant thinking history (#1000).
+	strip-thinking-history #true
 	thinking-mode "effort"
 	// Same runtime as local Ollama: a trailing assistant turn is continued verbatim.
 	supports-assistant-prefill #true
@@ -20,7 +22,12 @@ provider "ollama-cloud" {
 		thinking-efforts "minimal" "low" "medium" "high"
 	}
 	// residue: taxonomy ranks and exact globs do not isolate these models.
+	// #9693 verifies thinking history for these GLM-5.2 wire IDs only.
 	models "glm-5.2" {
+		strip-thinking-history #false
 		thinking-efforts "high" "max"
 	}
+	models "glm-5.2:cloud" {
+		strip-thinking-history #false
+	}
 }

package/src/compat/rules/providers/openai-codex.kdl

@@ -4,6 +4,11 @@ provider "openai-codex" {
 	default-model "gpt-5.5"
 	env "OPENAI_CODEX_OAUTH_TOKEN"
 	dynamic-models-authoritative #true
+	// Codex `/models` `display_name` is a title-cased slug (`GPT-6.1-Sol`,
+	// `GPT-5.6-Luna`) while the generator names bundled rows from the shared
+	// reference (`GPT-6.1 Sol`); without this the composer chip flips spelling
+	// once live discovery lands.
+	keep-bundled-names #true
 	kind-apis {
 		image "openai-codex-responses"
 	}

package/src/compat/rules/providers/openrouter.kdl

@@ -119,6 +119,9 @@ provider "openrouter" {
 	models "~typesafe/*" "typesafe/*" {
 		kind "judge"
 	}
+	models "~typesafe/jev-latest" "typesafe/jev-1.13" "typesafe/jev-1.14" {
+		input-modalities "text" "image"
+	}
 	models "cohere/rerank-v3.5" {
 		kind "rerank"
 	}
@@ -151,6 +154,13 @@ provider "openrouter" {
 			allows-synthetic-reasoning-content-for-tool-calls #false
 		}
 	}
+	// OpenRouter forwards these hosted Muse SKUs to Meta's backend, which
+	// accepts only tool_choice "auto" and 400s on "none" (see providers/meta.kdl).
+	// Other Muse models, including muse-spark-open, retain discovered tool_choice.
+	models "meta/muse-spark-1.1" "meta/muse-spark-1.2" "meta/muse-spark-1.2-contributor" \
+		"meta/muse-spark-1.3" "meta/muse-spark-1.3-contributor" "meta/muse-wonder" {
+		supports-tool-choice #false
+	}
 	class "minimax" {
 		family "m3" {
 			thinking-efforts "minimal" "low" "medium" "high"

package/src/compat/rules/providers/typesafe.kdl

@@ -14,6 +14,9 @@ provider "typesafe" {
 			supports-tools #false
 		}
 	}
+	models "jev-latest" "jev-preview" "jev-1.13.0" "jev-1.14.0" {
+		input-modalities "text" "image"
+	}
 
 	models "*" {
 		kind "judge"

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

@@ -226,9 +226,10 @@ behavior {
 	// effort ladder that lists `/responses` therefore rides Responses, so
 	// thinking-off actually disables reasoning instead of clamping to the
 	// lowest effort. Ladder ids without `/responses` (deepseek-v4-flash-fast,
-	// gemini-3.7-flash, Qwen3.8-Flash, Qwen3.8-Max-0902, hy4-preview,
-	// space-bunny-alpha), the Qwen3.8 ids (chat already disables them through
-	// `enable_thinking: false`), and non-reasoning ids stay on chat completions.
+	// gemini-3.7-flash, glm-5.3-flash since 2026-10-14, Qwen3.8-Flash,
+	// Qwen3.8-Max-0902, hy4-preview, space-bunny-alpha), the Qwen3.8 ids (chat
+	// already disables them through `enable_thinking: false`), and
+	// non-reasoning ids stay on chat completions.
 	api-routes provider="commandcode" default="openai-completions" {
 		route "anthropic-messages" prefix="claude-"
 		route "openai-responses" exact="gpt-5.3-codex" exact="gpt-5.4" exact="gpt-5.4-mini" exact="gpt-5.5" \
@@ -237,7 +238,7 @@ behavior {
 			exact="deepseek/deepseek-v4-flash" exact="deepseek/deepseek-v4-flash-vision-exp" \
 			exact="deepseek/deepseek-v4-pro" exact="deepseek/deepseek-v4.1-flash" \
 			exact="deepseek/deepseek-v4.1-flash-fast" exact="moonshotai/Kimi-K3" \
-			exact="z-ai/glm-5.3-flash" exact="z-ai/glm-5.3-flashx" exact="zai-org/GLM-5.2" \
+			exact="z-ai/glm-5.3-flashx" exact="zai-org/GLM-5.2" \
 			exact="zai-org/GLM-5.3" exact="google/gemini-3.1-flash-lite" exact="google/gemini-3.5-flash" \
 			exact="google/gemini-3.5-flash-lite" exact="google/gemini-3.6-flash" \
 			exact="google/gemini-3.8-flash" exact="MiniMaxAI/MiniMax-M3" exact="stepfun/Step-5-Preview" \
@@ -262,6 +263,8 @@ behavior {
 	// at /responses (#10610, #12030). The Muse prefix covers gateway-first
 	// revisions; exact ids stay as cache-migration identity.
 	api-routes provider="opencode-zen" {
+		// Paid Argon is served only on Zen's native Google endpoint (#15977).
+		route "google-generative-ai" exact="gemini-4-argon"
 		route "anthropic-messages" exact="union-alpha"
 		route "openai-responses" prefix="muse-spark-" exact="muse-spark-1.3-contributor-free" exact="gpt-6-astra"
 		route "openai-completions" exact="minimax-m3" exact="minimax-m3-free"

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

@@ -16,6 +16,8 @@ class "openai" {
 
 	family "gpt" glob="gpt-*"
 	family "chatgpt" glob="chatgpt-*"
+	family "gpt-image" glob="gpt-image-*"
+	family "gpt-image" glob="chatgpt-image-*"
 	family "o-series" glob="o1"
 	family "o-series" glob="o1-*"
 	family "o-series" glob="o1.*"