@oh-my-pi/pi-catalog 18.11.0 → 18.11.1

package/CHANGELOG.md

@@ -2,6 +2,17 @@
 
 ## [Unreleased]
 
+## [18.11.1] - 2026-10-15
+
+### Fixed
+
+- Enabled long prompt-cache retention (`ttl: "1h"`) for Claude models on `google-vertex` ([#15505](https://github.com/can1357/oh-my-pi/pull/15505) by [@hpost](https://github.com/hpost))
+- Fixed Gemini 4 Argon on Google Vertex and the Gemini API failing every request with "Thinking budget is not supported for this model": Gemini 4 now sends `thinkingLevel` instead of `thinkingBudget`, and Argon offers only the low, medium, and high levels Google accepts ([#15643](https://github.com/can1357/oh-my-pi/pull/15643) by [@kevcube](https://github.com/kevcube))
+- Fixed the Antigravity model selector offering roster rows the backend marks disabled (`disabled: true`, e.g. `gemini-4-argon-disabled` on accounts without Argon access), which answered `Cloud Code Assist API error (404): Requested entity was not found.`; discovery now skips them the way it already skips `isInternal` rows ([#15684](https://github.com/can1357/oh-my-pi/pull/15684) by [@CharleoYang](https://github.com/CharleoYang))
+- Removed Google Antigravity's Gemini 3.5 Flash and its `gemini-3-flash` alias from the model list; Antigravity retired them and answers every request with a retirement notice instead of a reply ([#15445](https://github.com/can1357/oh-my-pi/pull/15445) by [@will-bogusz](https://github.com/will-bogusz))
+- Fixed `xai-oauth` Grok 4.20 models (`grok-4.20-0309-reasoning`, `-non-reasoning`, `grok-4.20-multi-agent-0309`) advertising a 2M context window when the SuperGrok API rejects prompts over 1M; auto-compaction and context usage now work from the 1M window ([#15443](https://github.com/can1357/oh-my-pi/pull/15443) by [@will-bogusz](https://github.com/will-bogusz))
+- Fixed trailing-dot `localhost.` and `*.local.` custom endpoints losing local-backend compatibility settings ([#15798](https://github.com/can1357/oh-my-pi/issues/15798)).
+
 ## [18.11.0] - 2026-10-14
 
 ### Added

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

@@ -18,6 +18,13 @@ export interface AntigravityDiscoveryApiModel {
     apiProvider?: string;
     modelProvider?: string;
     isInternal?: boolean;
+    /**
+     * Set for roster rows the backend lists but refuses to serve on this
+     * account (e.g. `gemini-4-argon-disabled`, `tagTitle: "Unavailable"`,
+     * `model: MODEL_PLACEHOLDER_M421`); a request against such an id answers
+     * `Cloud Code Assist API error (404): Requested entity was not found.`
+     */
+    disabled?: boolean;
     supportsVideo?: boolean;
 }
 /**

package/dist/types/types.d.ts

@@ -1230,6 +1230,8 @@ export interface Model<TApi extends Api = Api> {
     requiresGlyphTokenization?: boolean;
     /** Whether this model requires Cursor's tool-schema combiner projection. */
     requiresCursorToolSchemaProjection?: boolean;
+    /** Whether advertised tool schemas occupy every model request's context (false for deferred catalogs). */
+    toolSchemasInContext?: boolean;
     /** Whether this model requires tool-result images hoisted into sibling user content blocks. */
     requiresToolResultImageHoisting?: boolean;
     /**
@@ -1493,7 +1495,7 @@ export interface Model<TApi extends Api = Api> {
  * vocabulary of `buildModel`. Identical to `Model` except `compat` carries the
  * sparse override shape and nothing is resolved yet.
  */
-export interface ModelSpec<TApi extends Api = Api> extends Omit<Model<TApi>, "compat" | "identity" | "compatConfig" | "requiresGlyphTokenization" | "requiresCursorToolSchemaProjection" | "requiresToolResultImageHoisting" | "supportsAssistantPrefill" | "supportsComputerUseConfig"> {
+export interface ModelSpec<TApi extends Api = Api> extends Omit<Model<TApi>, "compat" | "identity" | "compatConfig" | "requiresGlyphTokenization" | "requiresCursorToolSchemaProjection" | "toolSchemasInContext" | "requiresToolResultImageHoisting" | "supportsAssistantPrefill" | "supportsComputerUseConfig"> {
     /** Sparse compatibility overrides; resolved into `Model.compat` by `buildModel`. */
     compat?: CompatConfigOf<TApi>;
 }

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@oh-my-pi/pi-catalog",
-	"version": "18.11.0",
+	"version": "18.11.1",
 	"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": "18.11.0",
-		"@oh-my-pi/pi-utils": "18.11.0"
+		"@oh-my-pi/omptype": "18.11.1",
+		"@oh-my-pi/pi-utils": "18.11.1"
 	},
 	"devDependencies": {
 		"@bgotink/kdl": "0.4.0",
-		"@oh-my-pi/pi-ai": "18.11.0",
+		"@oh-my-pi/pi-ai": "18.11.1",
 		"@types/bun": "^1.3.14"
 	},
 	"engines": {

package/src/build.ts

@@ -115,6 +115,11 @@ function applyCatalogAssignments<TApi extends Api>(model: Model<TApi>, catalog:
 	} else {
 		delete model.requiresCursorToolSchemaProjection;
 	}
+	if (catalog.toolSchemasInContext === false) {
+		model.toolSchemasInContext = false;
+	} else {
+		delete model.toolSchemasInContext;
+	}
 	const requiresToolResultImageHoisting = catalog.requiresToolResultImageHoisting;
 	if (requiresToolResultImageHoisting === true) {
 		model.requiresToolResultImageHoisting = true;

package/src/compat/axes.ts

@@ -340,6 +340,7 @@ export const AXES: Readonly<Record<string, AxisDef>> = {
 		values: ["freeform", "function"],
 	},
 	"requires-native-tools": { key: "requiresNativeTools", set: "catalog", shape: "scalar", values: [true, false] },
+	"tool-schemas-in-context": { key: "toolSchemasInContext", set: "catalog", shape: "scalar", values: [true, false] },
 	"requires-tool-free-history-for-tool-opt-out": {
 		key: "requiresToolFreeHistoryForToolOptOut",
 		set: "catalog",

package/src/compat/resolve.ts

@@ -276,6 +276,7 @@ function hasLocalLoopbackBaseUrl(baseUrl: string | undefined): boolean {
 	let hostname: string;
 	try {
 		hostname = new URL(baseUrl).hostname.toLowerCase();
+		if (hostname.endsWith(".")) hostname = hostname.slice(0, -1);
 	} catch {
 		return false;
 	}
@@ -1039,7 +1040,8 @@ function defaultThinkingMode<TApi extends Api>(spec: ModelSpec<TApi>, facts: Ide
 		case "google-generative-ai":
 		case "google-gemini-cli":
 		case "google-vertex":
-			return facts.is("gemini") && facts.revMajor() === 3 ? "google-level" : "budget";
+			// Gemini 3+ `google-level` is a `thinking-mode` rule in classes/gemini.kdl.
+			return "budget";
 		case "anthropic-messages":
 			if (facts.is("minimax") && facts.family("m2", "m3")) return "anthropic-adaptive";
 			if (facts.is("glm") && facts.revGte("5.2") && (spec.provider === "umans" || spec.provider === "zai")) {

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

@@ -66,13 +66,15 @@ class "gemini" {
 			}
 		}
 	}
-	// Gemini 3.7+ Flash native thinkingLevel does not accept `minimal`
-	// (400 THINKING_LEVEL_MINIMAL; observed on 3.7 and 3.8). Only the direct
+	// Gemini 3.7+ Flash and Gemini 4 Argon native thinkingLevel does not accept
+	// `minimal` (400 THINKING_LEVEL_MINIMAL; observed on Flash 3.7 and 3.8, and
+	// documented for Argon and Argon Cyber Permissive). Only the direct
 	// google-level transports emit `thinkingLevel` on the wire, so the tier is
 	// dropped there; budget (anthropic-messages) and reasoning_effort resellers
 	// never send the rejected value and keep the four-tier ladder. The collapsed
 	// google-antigravity / google-gemini-cli variants remap minimal -> LOW and
 	// carry their own explicit ladder, so they are unaffected.
+	// https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/4-argon
 	// https://ai.google.dev/gemini-api/docs/generate-content/thinking#thinking-levels
 	on "google" "google-vertex" "opencode-zen" {
 		family "flash" {
@@ -80,6 +82,13 @@ class "gemini" {
 				thinking-efforts "low" "medium" "high"
 			}
 		}
+		// Exact ids: a revision-wide rule would tie with the Pro low/high ladder.
+		// Gemini API discovery has no bundled Argon reference and infers
+		// `reasoning: false` from the id; repair it so the ladder is selectable.
+		models "gemini-4-argon" "gemini-4-argon-cyber-permissive" {
+			thinking-upgrade-neutral #true
+			thinking-efforts "low" "medium" "high"
+		}
 	}
 	// Gemini 4 Argon's introductory list price ($2/$10, $0.10 cached) ends
 	// 2026-11-30; from 2026-12-01 Google bills $4/$20, $0.20 cached, on the
@@ -116,6 +125,13 @@ class "gemini" {
 		// Replaces the Gemini 3+ mandatory-reasoning fallback.
 		thinking-requires-effort #true
 	}
+	// Gemini 3+ takes `thinkingLevel` on the native Google transports; Gemini 4
+	// rejects `thinkingBudget` outright (400 "Use thinking level instead").
+	on-api "google-generative-ai" "google-gemini-cli" "google-vertex" {
+		revision ">=3" {
+			thinking-mode "google-level"
+		}
+	}
 	// Public Gemini requires the bypass sentinel on every unsigned call and
 	// supports function-part ids. Scope by adapter so custom providers using
 	// the same wire protocol inherit both contracts. The Cloud Code Assist

package/src/compat/rules/providers/google-vertex.kdl

@@ -28,6 +28,10 @@ provider "google-vertex" {
 		}
 	}
 	class "anthropic" {
+		// Vertex rawPredict honors `cache_control.ttl: "1h"` (writes bill to
+		// `ephemeral_1h_input_tokens`, entries outlive 5m) without any beta; the
+		// `extended-cache-ttl-2025-04-11` beta is rejected in `anthropic_beta`.
+		supports-long-cache-retention #true
 		revision ">=4.5 <4.6" {
 			thinking-efforts "minimal" "low" "medium" "high" "xhigh"
 		}

package/src/compat/rules/providers/native-tools.kdl

@@ -1,4 +1,7 @@
 // Cursor always exposes server-owned native operations, even with an empty tool catalog.
 on-api "cursor-agent" {
 	requires-native-tools #true
+	// MCP schemas live in Cursor's server-side catalog and enter model context
+	// only when dynamically selected, not as a fixed cost on every turn.
+	tool-schemas-in-context #false
 }

package/src/compat/rules/providers/xai-oauth.kdl

@@ -15,8 +15,9 @@ provider "xai-oauth" {
 
 	// Source of truth for the xai-oauth chat picker. Declaration order is headline order.
 	// Context windows from hermes-agent/agent/model_metadata.py:205-220
-	// ("Values sourced from models.dev (2026-04)"). grok-build is xAI's
-	// coding-fine-tuned chat model; 512K context per user spec (2026-05-17).
+	// ("Values sourced from models.dev (2026-04)") unless a row says otherwise.
+	// grok-build is xAI's coding-fine-tuned chat model; 512K context per user
+	// spec (2026-05-17).
 	//
 	// xai-oauth's /v1/models exposes no per-request output limit on the OAuth
 	// (Grok Build / SuperGrok) surface, so maxTokens mirrors contextWindow. The
@@ -68,25 +69,27 @@ provider "xai-oauth" {
 			cost input=0 output=0 cache-read=0 cache-write=0
 			limits context=500000 max-tokens=500000
 		}
-		// Text-only per the bundled catalog.
+		// grok-4.20-*-0309: api.x.ai rejects prompts over 1M tokens on this surface
+		// ([input_too_large] "... > 1000000 tokens", all three ids, 2026-10-12).
+		// Multi-agent is text-only per the bundled catalog.
 		model "grok-4.20-multi-agent-0309" name="Grok 4.20 (Multi-Agent)" {
 			reasoning #true
 			input "text"
 			cost input=0 output=0 cache-read=0 cache-write=0
-			limits context=2000000 max-tokens=2000000
+			limits context=1000000 max-tokens=1000000
 		}
 		model "grok-4.20-0309-reasoning" name="Grok 4.20 (Reasoning)" {
 			reasoning #true
 			input "text" "image"
 			cost input=0 output=0 cache-read=0 cache-write=0
-			limits context=2000000 max-tokens=2000000
+			limits context=1000000 max-tokens=1000000
 			supports-reasoning-effort #false
 		}
 		model "grok-4.20-0309-non-reasoning" name="Grok 4.20 (Non-Reasoning)" {
 			reasoning #false
 			input "text" "image"
 			cost input=0 output=0 cache-read=0 cache-write=0
-			limits context=2000000 max-tokens=2000000
+			limits context=1000000 max-tokens=1000000
 		}
 		// Cursor's "Composer 2.5 Fast" exposed via SuperGrok: non-reasoning,
 		// text-only, 200K context (mirrors Cursor's composer-* catalog entries).

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

@@ -173,6 +173,10 @@ behavior {
 	// serve; the dedicated tool surfaces route them instead.
 	exclude-models provider="xai-oauth" exact="grok-imagine-image-quality" exact="grok-imagine-image-2.0" \
 		prefix="grok-imagine-video" prefix="grok-stt-" prefix="grok-voice-"
+	// Antigravity still advertises the Gemini 3.5 Flash wire ids, but every
+	// request to them answers with a retirement notice instead of a reply.
+	// `gemini-3-flash` is the family's recycled alias and routes to the same ids.
+	exclude-models provider="google-antigravity" exact="gemini-3.5-flash" exact="gemini-3-flash"
 	// Meta's /v1/models lists image generation and transcription SKUs beside
 	// the Muse Spark chat models, and SAM segmentation beside them.
 	exclude-models provider="meta" prefix="muse-image-" prefix="muse-voice-" prefix="sam-"
@@ -349,7 +353,6 @@ behavior {
 	// Google); price rows at their first-party equivalents, Vertex before
 	// Anthropic for Claude ids. Aliases map ids whose peer ships differently.
 	pricing-peer provider="google-antigravity" peers="google" "google-vertex" "anthropic" {
-		alias "gemini-3-flash" peer-id="gemini-3-flash-preview"
 		alias "gemini-3-pro" peer-id="gemini-3-pro-preview"
 		alias "gemini-3.1-pro" peer-id="gemini-3.1-pro-preview"
 		alias "claude-opus-4-5" peer-id="claude-opus-4-5@20251101"