@oh-my-pi/pi-catalog 16.3.12 → 16.3.14

package/CHANGELOG.md

@@ -2,6 +2,30 @@
 
 ## [Unreleased]
 
+## [16.3.14] - 2026-07-09
+
+### Added
+
+- Added support for GPT-5.6 (Luna, Sol, Terra) model variants
+- Enabled expanded five-tier reasoning effort scale (minimal to xhigh) for GPT-5.6 models
+- Added GPT-5.6 (Terra/Luna/Sol) support for the new `max` reasoning tier: on wire-effort APIs (OpenAI Responses, Codex, Azure, openai-compat/OpenRouter models that advertise reasoning) user efforts shift up one notch — `xhigh` sends `max`, `high` sends `xhigh` — mirroring the Claude Fable/Opus 4.7+ five-tier mapping, and the exposed ladder becomes `minimal..xhigh` with `minimal` reaching the native `low` tier. Devin's per-tier GPT-5.6 sibling rows now collapse into `gpt-5-6-{luna,sol,terra}` logical models with the same shifted routing (`xhigh` → `-max`), plus `-fast` families that keep the direct `low..xhigh` `-priority` scale since Devin serves no `-max-priority` tier.
+
+## [16.3.13] - 2026-07-09
+
+### Added
+
+- Added support for Grok 4.5 across multiple providers
+- Added support for GPT-5.6 series models (Luna, Sol, Terra)
+- Added Aion 3.0 and 3.0 Mini models
+- Added Kuaishou KAT-Coder v2.5 models
+- Added Nex-N2-Mini and SWE-1.7 series models
+- Added Hy3 models and free variants
+
+### Changed
+
+- Updated cost and token configurations for various models across providers
+- Renamed several models for consistency (e.g., MiniMax M3, Gemma 4 31B, Qwen variants)
+
 ## [16.3.12] - 2026-07-08
 
 ### Fixed

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

@@ -24,7 +24,7 @@ export interface ModelManagerOptions<TApi extends Api = Api, TModelsDevPayload =
     cacheDbPath?: string;
     /** Optional provider id override for cache namespacing. Defaults to providerId. */
     cacheProviderId?: string;
-    /** Maximum cache age in milliseconds before considered stale. Default: 24h. */
+    /** Maximum cache age in milliseconds before considered stale. Default: 2h (`DEFAULT_CACHE_TTL_MS`). */
     cacheTtlMs?: number;
     /** When true, a successful dynamic fetch is the complete provider catalog and prunes static-only models. */
     dynamicModelsAuthoritative?: boolean;

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

@@ -14,12 +14,14 @@ import type { Api, CompatOf, Model, ModelSpec, ThinkingConfig } from "./types.js
  */
 type ApiModel<TApi extends Api = Api> = ModelSpec<TApi> | Model<TApi>;
 /**
- * Effort → wire-value map for the 5-tier adaptive scale (Opus 4.7+ and
- * Fable/Mythos 5 on the Messages API). User-facing efforts shift up one notch
- * so the top tier reaches the genuine "max" and "high" lands on Anthropic's
- * recommended "xhigh" coding/agentic default.
+ * Effort → wire-value map for a shifted five-tier scale (`low..max`):
+ * user-facing efforts shift up one notch so the top tier reaches the genuine
+ * "max" and "high" lands on the recommended "xhigh" coding/agentic default.
+ * Used by Anthropic adaptive models with a real xhigh tier (Opus 4.7+ and
+ * Fable/Mythos 5 on the Messages API) and by GPT-5.6+ wire-effort models,
+ * which expose the same genuine `max` tier above `xhigh`.
  */
-export declare const ANTHROPIC_ADAPTIVE_EFFORT_MAP_5_TIER: Readonly<Partial<Record<Effort, string>>>;
+export declare const SHIFTED_FIVE_TIER_EFFORT_MAP: Readonly<Partial<Record<Effort, string>>>;
 /**
  * Effort → wire-value map for the legacy 4-tier adaptive scale (Opus 4.6,
  * Sonnet 4.6+, and every adaptive model on Bedrock Converse). `low..high` pass

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@oh-my-pi/pi-catalog",
-	"version": "16.3.12",
+	"version": "16.3.14",
 	"description": "Model catalog for omp: bundled model database, provider discovery descriptors, model identity, classification, and equivalence",
 	"homepage": "https://omp.sh",
 	"author": "Can Boluk",
@@ -34,12 +34,12 @@
 	},
 	"dependencies": {
 		"@bufbuild/protobuf": "^2.12.0",
-		"@oh-my-pi/pi-utils": "16.3.12",
+		"@oh-my-pi/pi-utils": "16.3.14",
 		"arktype": "^2.2.0",
 		"zod": "^4"
 	},
 	"devDependencies": {
-		"@oh-my-pi/pi-ai": "16.3.12",
+		"@oh-my-pi/pi-ai": "16.3.14",
 		"@types/bun": "^1.3.14"
 	},
 	"engines": {

package/src/discovery/cursor.ts

@@ -13,6 +13,13 @@ const CURSOR_GET_USABLE_MODELS_PATH = "/agent.v1.AgentService/GetUsableModels";
 const DEFAULT_CONTEXT_WINDOW = 200_000;
 const DEFAULT_MAX_TOKENS = 64_000;
 
+/**
+ * Model-id families whose native catalogs (anthropic, openai/openai-codex,
+ * google) are multimodal. Cursor-only or text-only families (`composer-*`,
+ * `grok-code-*`) intentionally stay outside this pattern.
+ */
+const CURSOR_MULTIMODAL_ID_PATTERN = /claude|gemini|gpt-|codex/;
+
 const OptionalDisplayNameSchema = type("unknown").pipe(raw => (typeof raw === "string" ? raw : undefined));
 const CursorAliasesSchema = type("unknown").pipe(raw => {
 	if (Array.isArray(raw)) {
@@ -292,7 +299,7 @@ function normalizeCursorModel(
 		provider: "cursor",
 		baseUrl: baseUrlOverride ?? CURSOR_DEFAULT_BASE_URL,
 		reasoning,
-		input: ["text"],
+		input: inferInputFromCursorId(id),
 		cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
 		contextWindow: DEFAULT_CONTEXT_WINDOW,
 		maxTokens: DEFAULT_MAX_TOKENS,
@@ -312,3 +319,18 @@ function pickModelDisplayName(model: CursorModelDetailsValue, fallbackId: string
 	}
 	return fallbackId;
 }
+
+/**
+ * Infers input modalities for Cursor models without a bundled reference.
+ *
+ * `GetUsableModels` carries no per-model modality metadata, so classification
+ * falls back to the model family: families that are multimodal in OMP's own
+ * native catalogs accept images, everything else stays text-only. Mirrors
+ * `inferInputFromGeminiId` in ./gemini.ts.
+ */
+function inferInputFromCursorId(id: string): ("text" | "image")[] {
+	if (CURSOR_MULTIMODAL_ID_PATTERN.test(id.toLowerCase())) {
+		return ["text", "image"];
+	}
+	return ["text"];
+}

package/src/model-manager.ts

@@ -35,7 +35,7 @@ export interface ModelManagerOptions<TApi extends Api = Api, TModelsDevPayload =
 	cacheDbPath?: string;
 	/** Optional provider id override for cache namespacing. Defaults to providerId. */
 	cacheProviderId?: string;
-	/** Maximum cache age in milliseconds before considered stale. Default: 24h. */
+	/** Maximum cache age in milliseconds before considered stale. Default: 2h (`DEFAULT_CACHE_TTL_MS`). */
 	cacheTtlMs?: number;
 	/** When true, a successful dynamic fetch is the complete provider catalog and prunes static-only models. */
 	dynamicModelsAuthoritative?: boolean;

package/src/model-thinking.ts

@@ -17,6 +17,7 @@ import {
 	type ParsedModel,
 	parseAnthropicModel,
 	parseKnownModel,
+	parseOpenAIModel,
 	semverEqual,
 	semverGte,
 } from "./identity/classify";
@@ -101,12 +102,14 @@ const MIMO_REASONING_EFFORT_MAP: Readonly<EffortMap> = {
 };
 
 /**
- * Effort → wire-value map for the 5-tier adaptive scale (Opus 4.7+ and
- * Fable/Mythos 5 on the Messages API). User-facing efforts shift up one notch
- * so the top tier reaches the genuine "max" and "high" lands on Anthropic's
- * recommended "xhigh" coding/agentic default.
+ * Effort → wire-value map for a shifted five-tier scale (`low..max`):
+ * user-facing efforts shift up one notch so the top tier reaches the genuine
+ * "max" and "high" lands on the recommended "xhigh" coding/agentic default.
+ * Used by Anthropic adaptive models with a real xhigh tier (Opus 4.7+ and
+ * Fable/Mythos 5 on the Messages API) and by GPT-5.6+ wire-effort models,
+ * which expose the same genuine `max` tier above `xhigh`.
  */
-export const ANTHROPIC_ADAPTIVE_EFFORT_MAP_5_TIER: Readonly<Partial<Record<Effort, string>>> = {
+export const SHIFTED_FIVE_TIER_EFFORT_MAP: Readonly<Partial<Record<Effort, string>>> = {
 	[Effort.Minimal]: "low",
 	[Effort.Low]: "medium",
 	[Effort.Medium]: "high",
@@ -295,6 +298,27 @@ function isOpenAICompatReasoningApi(api: Api): boolean {
 	return api === "openai-completions" || api === "openrouter";
 }
 
+/**
+ * GPT-5.6+ addressed through a wire `reasoning.effort`/`reasoning_effort`
+ * field, where the shifted five-tier map applies. Devin (`devin-agent`)
+ * selects effort by routing to per-tier sibling model ids instead and must
+ * stay unmapped.
+ */
+function isGpt56PlusWireEffortModel<TApi extends Api>(spec: ModelSpec<TApi>): boolean {
+	switch (spec.api) {
+		case "openai-responses":
+		case "openai-codex-responses":
+		case "azure-openai-responses":
+		case "openai-completions":
+		case "openrouter":
+			break;
+		default:
+			return false;
+	}
+	const parsed = parseOpenAIModel(bareModelId(spec.id));
+	return parsed !== null && semverGte(parsed.version, "5.6");
+}
+
 function getModelDefinedEfforts<TApi extends Api>(
 	spec: ModelSpec<TApi>,
 	compat: CompatOf<TApi>,
@@ -313,6 +337,12 @@ function getModelDefinedEfforts<TApi extends Api>(
 	if (isSakanaFuguReasoningModel(spec)) {
 		return FUGU_REASONING_EFFORTS;
 	}
+	if (isGpt56PlusWireEffortModel(spec)) {
+		// Normalize stale baked/discovered `low..xhigh` surfaces to the full
+		// five-tier ladder so the shifted map keeps the native `low` tier
+		// reachable (user `minimal`).
+		return DEFAULT_REASONING_EFFORTS_WITH_XHIGH;
+	}
 	return isOpenAICompatReasoningApi(spec.api) &&
 		(isMinimaxM2FamilyModelId(spec.id) ||
 			isOpenAIGptOssModelId(spec.id) ||
@@ -373,7 +403,7 @@ function inferDetectedEffortMap<TApi extends Api>(
 			return MINIMAX_ANTHROPIC_ADAPTIVE_EFFORT_MAP;
 		}
 		return anthropicModelHasRealXHighEffort(spec, parsedModel)
-			? ANTHROPIC_ADAPTIVE_EFFORT_MAP_5_TIER
+			? SHIFTED_FIVE_TIER_EFFORT_MAP
 			: ANTHROPIC_ADAPTIVE_EFFORT_MAP_4_TIER;
 	}
 	// GLM-5.2 coding SKUs accept `reasoning_effort`, but the effort dialect is
@@ -397,6 +427,9 @@ function inferDetectedEffortMap<TApi extends Api>(
 	if (isSakanaFuguReasoningModel(spec)) {
 		return FUGU_REASONING_EFFORT_MAP;
 	}
+	if (isGpt56PlusWireEffortModel(spec)) {
+		return SHIFTED_FIVE_TIER_EFFORT_MAP;
+	}
 	if (!isOpenAICompatReasoningApi(spec.api)) {
 		return undefined;
 	}
@@ -446,7 +479,7 @@ function getOpenRouterAnthropicReasoningEffortMap(modelId: string): EffortMap |
 	if (!isAnthropicAdaptiveGenAtLeast(parsed, "4.6")) return undefined;
 
 	const hasRealXHigh = isAnthropicAdaptiveGenAtLeast(parsed, "4.7");
-	return hasRealXHigh ? ANTHROPIC_ADAPTIVE_EFFORT_MAP_5_TIER : ANTHROPIC_ADAPTIVE_EFFORT_MAP_4_TIER;
+	return hasRealXHigh ? SHIFTED_FIVE_TIER_EFFORT_MAP : ANTHROPIC_ADAPTIVE_EFFORT_MAP_4_TIER;
 }
 
 function inferSupportedEfforts<TApi extends Api>(
@@ -474,6 +507,11 @@ function inferOpenAISupportedEfforts(model: OpenAIModel): readonly Effort[] {
 	if (model.variant === "codex-mini" && semverEqual(model.version, "5.1")) {
 		return GPT_5_1_CODEX_MINI_EFFORTS;
 	}
+	// 5.6+ exposes the full five-tier ladder: the shifted wire map spans
+	// low..max, with user `minimal` reaching the native `low` tier.
+	if (semverGte(model.version, "5.6")) {
+		return DEFAULT_REASONING_EFFORTS_WITH_XHIGH;
+	}
 	if (semverGte(model.version, "5.2")) {
 		return GPT_5_2_PLUS_EFFORTS;
 	}