@oh-my-pi/pi-catalog 15.12.3 → 15.13.0

package/CHANGELOG.md

@@ -2,90 +2,20 @@
 
 ## [Unreleased]
 
-## [15.11.8] - 2026-06-12
-
-### Fixed
-
-- Fixed Antigravity `gemini-3.1-pro --thinking high` failing with `Cloud Code Assist API error (400): Request contains an invalid argument.` — the upstream `gemini-3.1-pro-high` deployment rejects every `streamGenerateContent` request on both CCA endpoints while discovery still advertises it. High effort now routes to `gemini-pro-agent` (the same "Gemini 3.1 Pro (High)" model, verified accepting the identical request body), and the model-cache fingerprint version was bumped (`merge-v2` → `merge-v3`) so existing fresh caches refetch discovery and pick up the corrected routing immediately.
-
-## [15.11.7] - 2026-06-12
 ### Added
 
+- Added `modelFamilyToken(modelId)` to `@oh-my-pi/pi-catalog/identity`: a coarse vendor-lineage token (`anthropic`/`openai`/`gemini`/`kimi`/…) for "are two models the same family?" comparisons, backed by `parseKnownModel` canonical-id normalization. Opaque and comparison-only; kind/variant collapsed onto the vendor token ([#2406](https://github.com/can1357/oh-my-pi/issues/2406))
+- Added GLM-5.2 to the bundled zai (GLM Coding Plan) catalog as the selectable 1M served model.
+- Added bundled Fireworks models `deepseek-v4-flash`, `kimi-k2.7-code`, `minimax-m2.5`, `minimax-m3`, `nemotron-3-ultra-nvfp4`, `qwen3.6-plus`, and `qwen3.7-plus`
+- Changed
 - Added effort-tier variant collapsing (`variant-collapse`): providers that expose one logical model as several effort/thinking-suffixed upstream ids (Antigravity CCA `gemini-3.5-flash-extra-low`/`-low`/`gemini-3-flash-agent`, `gemini-3[.1]-pro-low|high`, `claude-*[-thinking]` pairs, `gpt-oss-120b-medium`) collapse into one logical entry carrying per-effort upstream routing in `thinking.effortRouting` (plus `thinking.suppressWhenOff` for Cloud Code Assist ids whose baked server default re-applies when `thinkingConfig` is omitted). Request-time code resolves the outbound id via `resolveWireModelId(model, effort)`; selection, caching, and usage attribution key on the logical id.
 - Added the automatic `X`/`X-thinking` pair rule (`deriveThinkingPairFamilies`): any provider's live bare/thinking twin collapses into the bare id, routing thinking-enabled requests to the `-thinking` backing id (trailing or infix token, so `kimi-k2-thinking-turbo` pairs with `kimi-k2-turbo`). Gated on same api and compatible pricing — all-zero cost rows count as unknown, while twins that both carry real, differing prices remain separate SKUs.
 - Added `collapseBuiltModelVariants` and wired collapsing at every materialization point — Antigravity discovery, the catalog generator, and the model-manager merge — so stale sources (old static beside collapsed dynamic results, mixed cache rows) converge on logical entries instead of unioning raw tier ids back into the catalog.
 - Added `thinking.requiresEffort`, baked for reasoning-only upstreams — Gemini 3.x (levels only, no off), Gemini 2.5 Pro (thinkingBudget floors at 128, rejects 0), OpenAI o-series, MiniMax M2, and thinking-variant SKUs (`*-thinking`/`*-reasoner`/`*-reasoning`, with a negation-aware token grammar so `non-thinking` ids never match). Identity derivation bakes it for new entries and `fillThinkingWireDefaults` backfills explicit/cached metadata; `minimumSupportedEffort` exposes the canonical floor. Pair-collapsed twins drop member flags (their off routes to the bare SKU), while identity re-flags pairs whose logical id is itself mandatory
-
-### Changed
-
-- Changed model display names to drop model-extrinsic decorations: gateway author prefixes (`OpenAI: …`, `Google: …`), `(latest)` alias markers, `(Antigravity)` provider attribution, price tiers (`($$$$)`), and promo/lifecycle tags (`(20% off)`, `(retires …)`). `cleanModelName` is applied in `buildModel` (covers live discovery and stale caches) and as a catalog-generator pass; Antigravity discovery no longer appends `(Antigravity)` to display names. Variant tags that map to distinct wire ids (`(Thinking)`, `(free)`, `(Fast)`, dates, regions) are preserved.
-- Changed the `google-antigravity` default model from `gemini-3-pro-high` to `gemini-3.1-pro`
-- Changed `gemini-2.5-flash-thinking` handling from discovery-denylist to collapsing into `gemini-2.5-flash` (thinking-enabled requests route to the `-thinking` backing id)
-- Bumped the model cache schema to v5 so rows predating effort-tier variant collapsing (raw `-low`/`-high`/`-thinking` member ids) are invalidated
-
-### Fixed
-
-- Fixed catalog generation to apply effort-tier variant collapsing before provider grouping to ensure collapsed model families are consistently materialized without being impacted by in-loop mutation
-- Fixed Kimi K2.6 OpenAI-compatible compat metadata to use a 300s stream watchdog floor, covering Fire Pass router ids as well as public `kimi-k2.6` ids so long reasoning starts do not hit the generic first-event timeout ([#2366](https://github.com/can1357/oh-my-pi/issues/2366)).
-
-## [15.11.4] - 2026-06-12
-
-### Fixed
-
-- Fixed MiniMax M2-family and OpenAI gpt-oss model metadata so OpenAI-compatible catalog entries declare only `low|medium|high` thinking efforts. Their upstreams reject `minimal`, `xhigh`, and Fireworks' `minimal → none` wire mapping, so `fireworks/minimax-m2.7` as the smol auto-thinking classifier model 400ed on every turn. OpenAI-compatible provider effort maps (`Groq qwen/qwen3-32b`, DeepSeek-family, OpenRouter Anthropic adaptive, Fireworks `minimal → none`) now bake into `thinking.effortMap` in catalog metadata instead of `buildOpenAICompat`, and request builders read that field directly. Regenerated `models.json` now makes `disableReasoning` choose `low` for those families while leaving GLM-5.x and other Fireworks models on the existing `minimal → none` path ([#2315](https://github.com/can1357/oh-my-pi/issues/2315)).
-### Added
-
 - Added `requiresJuiceZeroHack` Responses-API compat flag, resolved by `buildOpenAIResponsesCompat` from GPT-5-family model names and overridable via sparse model `compat` config. Replaces the request-time `model.name.startsWith("gpt-5")` sniff that gated the trailing `# Juice: 0 !important` no-reasoning developer item.
-
-## [15.11.3] - 2026-06-11
-### Added
-
 - Added `requestModelId` on `Model` to represent the upstream model id used when a catalog entry is a local variant
 - Added synthetic GitHub Copilot long-context model variants with `-1m` suffixes when tiered token pricing is advertised
-
-### Changed
-
-- Changed GitHub Copilot discovery to request `X-GitHub-Api-Version: 2026-06-01` from `api.githubcopilot.com`
-- Changed GitHub Copilot discovery to cap base model `contextWindow` to the default token tier and keep long-context access as the separate `-1m` model entry
-- Changed Copilot model mapping to omit non-chat `/models` entries and enable image input for models whose capabilities indicate vision support
-
-### Fixed
-
-- Fixed long-context variant pricing to use `billing.token_prices.long_context` rates instead of default model pricing
-- Fixed `mapModel` handling in OpenAI-compatible discovery so returning `null` now skips a model entry rather than falling back to defaults
-- Fixed model ID precedence so a real upstream Copilot model id is kept when it conflicts with a synthesized `-1m` variant
-
-## [15.11.1] - 2026-06-11
-
-### Fixed
-
-- Fixed NVIDIA NIM Qwen turns failing with `400 Validation: Unsupported parameter(s): enable_thinking`. NIM's chat-completions schema is `additionalProperties: false` and exposes thinking via the vLLM convention `chat_template_kwargs.enable_thinking`; `buildOpenAICompat` was sending top-level `enable_thinking` for every `qwen/*` id regardless of host. Registered `nvidia` as a known host (`integrate.api.nvidia.com`) and routed NVIDIA-hosted Qwen models to `thinkingFormat: "qwen-chat-template"` ([#2299](https://github.com/can1357/oh-my-pi/issues/2299)).
-- Fixed Moonshot/Kimi native OpenAI-compatible request metadata so Kimi K2 uses `max_tokens` and omits OpenAI-only `store`, restoring first-turn output with `MOONSHOT_API_KEY` ([#2289](https://github.com/can1357/oh-my-pi/issues/2289)).
-
-## [15.11.0] - 2026-06-10
-
-### Fixed
-
-- Fixed `buildModel` so malformed explicit thinking metadata without `efforts` is treated as sparse input and inferred instead of crashing during model resolution ([#2251](https://github.com/can1357/oh-my-pi/issues/2251)).
-
-## [15.10.12] - 2026-06-10
-
-### Added
-
 - Added `grok-composer-2.5-fast` (Cursor "Composer 2.5 Fast") to the xAI Grok OAuth (SuperGrok) catalog: non-reasoning, text-only, 200K context.
-
-### Changed
-
-- Set every xAI Grok OAuth (SuperGrok) curated model's max output tokens to mirror its context window (`grok-build`, `grok-4.3`, `grok-4.20-0309-{reasoning,non-reasoning}`, `grok-4.20-multi-agent-0309`, `grok-composer-2.5-fast`), replacing the `8888` `UNK_MAX_TOKENS` placeholder (and a stale `30000` on three grok-4.x entries). xAI's OAuth `/v1/models` reports no per-request output limit, so the curated catalog now owns `maxTokens` like `contextWindow`, deterministic on both the static-seed and online-overlay paths; the `openai-responses` wire still clamps the actual request to `OPENAI_MAX_OUTPUT_TOKENS` (64k).
-
-### Fixed
-
-- Excluded zero-cost `xai-oauth` subscription entries from the model reference indexes (`buildModelReferenceIndex`, `createReferenceResolver`), so their zero pricing and context-window-sized `maxTokens` cannot outrank paid/public Grok references when resolving custom-provider model identities.
-
-## [15.10.11] - 2026-06-10
-
-### Added
-
 - Added `hostMatchesUrl`, `modelMatchesHost`, and endpoint-shape helpers in the new `hosts` module for consistent provider/baseUrl matching
 - `buildModel(spec)` (`build.ts`) is now the single Model constructor: it materializes the fully-resolved compat record and canonical thinking metadata exactly once (compat first, thinking derived from identity + resolved compat), so `Model.compat` is a required, complete `CompatOf<TApi>` (`ResolvedOpenAICompat`/`ResolvedOpenAIResponsesCompat`/`ResolvedAnthropicCompat`) and request-path code reads fields with zero URL parsing and zero per-request allocation. Sparse user/config overrides live on the new `ModelSpec<TApi>` input shape and survive on `Model.compatConfig` for introspection.
 - Added `ResolvedAnthropicCompat.supportsSamplingParams` (Opus 4.7+/Fable/Mythos reject `temperature`/`top_p`/`top_k` with a 400), baked at build time from model identity so the request path stops re-parsing model ids.
@@ -97,6 +27,21 @@
 
 ### Changed
 
+- Changed catalog metadata to update a model’s per-token pricing to input 0.09 and output 0.18
+- Changed the same cataloged model’s maximum token limit from 384000 to 65536
+- Pinned zai `glm-5.2` to 1M context during catalog generation so endpoint discovery and older fallbacks cannot regress it to 200k.
+- Replaced the hand-maintained `zhipu-coding-plan` GLM reasoning allowlist and vision regex with a `parseGlmModel` family classifier in `identity/classify.ts` (variant + vision + version), surfaced as `isReasoningGlmModelId` / `isGlmVisionModelId`. Discovery now derives reasoning/vision capability from the GLM family instead of a per-id list, so newly-bumped integers (`glm-5.3`, `glm-6`, …) are covered automatically while `-flash`/`-preview` and the vision `…v` shape stay correctly classified.
+- Model `contextWindow`/`maxTokens` are now `number | null`; discovery emits `null` when a provider reports no limit, replacing the `222222`/`8888` (`UNK_CONTEXT_WINDOW`/`UNK_MAX_TOKENS`) sentinels (now removed). Bundled `models.json` unknown limits are `null`.
+- Changed the `github-copilot` model context window to `524288` tokens
+- Changed Fireworks model discovery to source the control-plane `List Models` API (`GET /v1/accounts/fireworks/models?filter=supports_serverless=true`) instead of the OpenAI-compatible `/v1/models` inference listing. The inference endpoint returns a sparse, account-specific subset that omits on-demand serverless models (e.g. `kimi-k2.7-code`), so newly published serverless models stayed invisible in the picker until hand-added to the bundled catalog. The control-plane catalog enumerates every serverless model with capability metadata (`supportsServerless`/`supportsTools`/`supportsImageInput`/`contextLength`/`displayName`), paginated and filtered to tool-capable `READY` entries, then merged with bundled/models.dev references — the Kimi K2 max-output clamp and DeepSeek V4 thinking-toggle strip are preserved, and unbundled models default to reasoning so `buildModel` derives the Fireworks effort map. New serverless releases now surface automatically with no catalog edits.
+- Changed model display names to drop model-extrinsic decorations: gateway author prefixes (`OpenAI: …`, `Google: …`), `(latest)` alias markers, `(Antigravity)` provider attribution, price tiers (`($$$$)`), and promo/lifecycle tags (`(20% off)`, `(retires …)`). `cleanModelName` is applied in `buildModel` (covers live discovery and stale caches) and as a catalog-generator pass; Antigravity discovery no longer appends `(Antigravity)` to display names. Variant tags that map to distinct wire ids (`(Thinking)`, `(free)`, `(Fast)`, dates, regions) are preserved.
+- Changed the `google-antigravity` default model from `gemini-3-pro-high` to `gemini-3.1-pro`
+- Changed `gemini-2.5-flash-thinking` handling from discovery-denylist to collapsing into `gemini-2.5-flash` (thinking-enabled requests route to the `-thinking` backing id)
+- Bumped the model cache schema to v5 so rows predating effort-tier variant collapsing (raw `-low`/`-high`/`-thinking` member ids) are invalidated
+- Changed GitHub Copilot discovery to request `X-GitHub-Api-Version: 2026-06-01` from `api.githubcopilot.com`
+- Changed GitHub Copilot discovery to cap base model `contextWindow` to the default token tier and keep long-context access as the separate `-1m` model entry
+- Changed Copilot model mapping to omit non-chat `/models` entries and enable image input for models whose capabilities indicate vision support
+- Set every xAI Grok OAuth (SuperGrok) curated model's max output tokens to mirror its context window (`grok-build`, `grok-4.3`, `grok-4.20-0309-{reasoning,non-reasoning}`, `grok-4.20-multi-agent-0309`, `grok-composer-2.5-fast`), replacing the `8888` `UNK_MAX_TOKENS` placeholder (and a stale `30000` on three grok-4.x entries). xAI's OAuth `/v1/models` reports no per-request output limit, so the curated catalog now owns `maxTokens` like `contextWindow`, deterministic on both the static-seed and online-overlay paths; the `openai-responses` wire still clamps the actual request to `OPENAI_MAX_OUTPUT_TOKENS` (64k).
 - Changed OpenAI compatibility detection to use shared host classifiers (`modelMatchesHost`/`hostMatchesUrl`) with normalized matching instead of raw URL substring checks
 - Changed `hostMatchesUrl`/`modelMatchesHost` usage in compatibility detection to reduce mismatches across case variants and provider alias hosts
 - Provider catalog entries now carry the runtime API-key env fallback as an ordered `envVars` list; `catalogDiscovery.envVars` became an optional generation-time override (only `cursor` and `vercel-ai-gateway` differ) and `PROVIDER_DESCRIPTORS` materializes the resolved list for `generate-models.ts`.
@@ -107,6 +52,25 @@
 
 ### Fixed
 
+- Fixed MiniMax-M3 catalog context for `minimax` and `minimax-cn` to report the documented 1M long-context tier instead of the upstream 512K pricing boundary ([#2576](https://github.com/can1357/oh-my-pi/issues/2576)).
+- Fixed OpenCode Go MiMo catalog metadata so title generation and other tool-enabled calls omit unsupported `tool_choice` instead of triggering provider 400s ([#2509](https://github.com/can1357/oh-my-pi/issues/2509)).
+- Fixed OpenCode Go `kimi-k2.7-code` catalog metadata so resolve-gate requests use automatic tool selection instead of Moonshot-rejected forced `tool_choice` ([#2546](https://github.com/can1357/oh-my-pi/issues/2546)).
+- Fixed Anthropic compat for the `github-copilot` host so `supportsEagerToolInputStreaming` defaults to `false` there, matching the Copilot proxy which rejects the per-tool `eager_input_streaming` field ([#2558](https://github.com/can1357/oh-my-pi/issues/2558)).
+- Scoped vLLM model cache validity to the discovery base URL so changed endpoints refetch immediately, and bounded built-in vLLM discovery requests with a timeout.
+- Filled missing `contextWindow` and `maxTokens` in generated `models.json` for proxy/reseller variants by inheriting limits from canonical-family and segment-reference models
+- Ignored zero-cost `x-ai` subscription entries as reference sources when backfilling limits so inflated values are not propagated
+- Fixed the model cache opening with `PRAGMA journal_mode=WAL` before `PRAGMA busy_timeout`, so concurrent omp startups could crash inside `getDb()` on `SQLITE_BUSY` during WAL recovery instead of waiting through the transient lock. The busy handler is now installed before the first lock-taking statement ([#2421](https://github.com/can1357/oh-my-pi/issues/2421)).
+- Fixed Antigravity `gemini-3.1-pro --thinking high` failing with `Cloud Code Assist API error (400): Request contains an invalid argument.` — the upstream `gemini-3.1-pro-high` deployment rejects every `streamGenerateContent` request on both CCA endpoints while discovery still advertises it. High effort now routes to `gemini-pro-agent` (the same "Gemini 3.1 Pro (High)" model, verified accepting the identical request body), and the model-cache fingerprint version was bumped (`merge-v2` → `merge-v3`) so existing fresh caches refetch discovery and pick up the corrected routing immediately.
+- Fixed catalog generation to apply effort-tier variant collapsing before provider grouping to ensure collapsed model families are consistently materialized without being impacted by in-loop mutation
+- Fixed Kimi K2.6 OpenAI-compatible compat metadata to use a 300s stream watchdog floor, covering Fire Pass router ids as well as public `kimi-k2.6` ids so long reasoning starts do not hit the generic first-event timeout ([#2366](https://github.com/can1357/oh-my-pi/issues/2366)).
+- Fixed MiniMax M2-family and OpenAI gpt-oss model metadata so OpenAI-compatible catalog entries declare only `low|medium|high` thinking efforts. Their upstreams reject `minimal`, `xhigh`, and Fireworks' `minimal → none` wire mapping, so `fireworks/minimax-m2.7` as the smol auto-thinking classifier model 400ed on every turn. OpenAI-compatible provider effort maps (`Groq qwen/qwen3-32b`, DeepSeek-family, OpenRouter Anthropic adaptive, Fireworks `minimal → none`) now bake into `thinking.effortMap` in catalog metadata instead of `buildOpenAICompat`, and request builders read that field directly. Regenerated `models.json` now makes `disableReasoning` choose `low` for those families while leaving GLM-5.x and other Fireworks models on the existing `minimal → none` path ([#2315](https://github.com/can1357/oh-my-pi/issues/2315)).
+- Fixed long-context variant pricing to use `billing.token_prices.long_context` rates instead of default model pricing
+- Fixed `mapModel` handling in OpenAI-compatible discovery so returning `null` now skips a model entry rather than falling back to defaults
+- Fixed model ID precedence so a real upstream Copilot model id is kept when it conflicts with a synthesized `-1m` variant
+- Fixed NVIDIA NIM Qwen turns failing with `400 Validation: Unsupported parameter(s): enable_thinking`. NIM's chat-completions schema is `additionalProperties: false` and exposes thinking via the vLLM convention `chat_template_kwargs.enable_thinking`; `buildOpenAICompat` was sending top-level `enable_thinking` for every `qwen/*` id regardless of host. Registered `nvidia` as a known host (`integrate.api.nvidia.com`) and routed NVIDIA-hosted Qwen models to `thinkingFormat: "qwen-chat-template"` ([#2299](https://github.com/can1357/oh-my-pi/issues/2299)).
+- Fixed Moonshot/Kimi native OpenAI-compatible request metadata so Kimi K2 uses `max_tokens` and omits OpenAI-only `store`, restoring first-turn output with `MOONSHOT_API_KEY` ([#2289](https://github.com/can1357/oh-my-pi/issues/2289)).
+- Fixed `buildModel` so malformed explicit thinking metadata without `efforts` is treated as sparse input and inferred instead of crashing during model resolution ([#2251](https://github.com/can1357/oh-my-pi/issues/2251)).
+- Excluded zero-cost `xai-oauth` subscription entries from the model reference indexes (`buildModelReferenceIndex`, `createReferenceResolver`), so their zero pricing and context-window-sized `maxTokens` cannot outrank paid/public Grok references when resolving custom-provider model identities.
 - Fixed Anthropic official-endpoint detection to require strict HTTPS hostname matching so non-official or lookalike URLs are no longer treated as official Anthropic hosts
 - Fixed Ollama Cloud dynamic discovery so same-id matches from other providers no longer supply context-window or max-output-token limits for discovered models.
 - Wired `@oh-my-pi/pi-catalog` into the release publish package list, tarball install smoke test, and root `bun generate-models` script.
@@ -115,4 +79,26 @@
 
 ### Removed
 
-- Removed the runtime enrichment layer: `enrichModelThinking` (and its non-enumerable memo-slot cache), `refreshModelThinking`, `modelOmitsReasoningEffort`, and the `model-thinking` re-exports of generator-only policies. Thinking metadata is resolved exactly once inside `buildModel`; runtime helpers (`getSupportedEfforts`, `clampThinkingLevelForModel`, `requireSupportedEffort`, the effort mappers) are pure field reads.
+- Removed the runtime enrichment layer: `enrichModelThinking` (and its non-enumerable memo-slot cache), `refreshModelThinking`, `modelOmitsReasoningEffort`, and the `model-thinking` re-exports of generator-only policies. Thinking metadata is resolved exactly once inside `buildModel`; runtime helpers (`getSupportedEfforts`, `clampThinkingLevelForModel`, `requireSupportedEffort`, the effort mappers) are pure field reads.
+
+## [15.13.0] - 2026-06-14
+
+## [15.12.6] - 2026-06-14
+
+## [15.12.4] - 2026-06-13
+
+## [15.11.8] - 2026-06-12
+
+## [15.11.7] - 2026-06-12
+
+## [15.11.4] - 2026-06-12
+
+## [15.11.3] - 2026-06-11
+
+## [15.11.1] - 2026-06-11
+
+## [15.11.0] - 2026-06-10
+
+## [15.10.12] - 2026-06-10
+
+## [15.10.11] - 2026-06-10

package/dist/types/identity/classify.d.ts

@@ -12,6 +12,7 @@ export type SemVer = {
 export type GeminiKind = "pro" | "flash";
 export type AnthropicKind = "opus" | "sonnet" | "fable" | "mythos";
 export type OpenAIVariant = "base" | "codex" | "codex-max" | "codex-mini" | "codex-spark" | "mini" | "max" | "nano";
+export type GlmVariant = "base" | "air" | "turbo" | "flash" | "flashx" | "preview";
 export interface GeminiModel {
     family: "gemini";
     kind: GeminiKind;
@@ -27,6 +28,14 @@ export interface OpenAIModel {
     variant: OpenAIVariant;
     version: SemVer;
 }
+export interface GlmModel {
+    family: "glm";
+    /** Suffix variant (`-air`, `-turbo`, `-flash`, `-flashx`, `-preview`); `base` when none. */
+    variant: GlmVariant;
+    /** Vision SKU — the `v` that attaches directly to the version (`glm-4v`, `glm-4.5v`). */
+    vision: boolean;
+    version: SemVer;
+}
 export interface UnknownModel {
     family: "unknown";
     id: string;
@@ -35,9 +44,18 @@ export type ParsedModel = GeminiModel | AnthropicModel | OpenAIModel | UnknownMo
 /** Strip a provider namespace prefix (`openai/gpt-5.4` → `gpt-5.4`). */
 export declare function bareModelId(modelId: string): string;
 export declare function parseKnownModel(modelId: string): ParsedModel;
-export declare function parseGeminiModel(modelId: string): GeminiModel | null;
-export declare function parseAnthropicModel(modelId: string): AnthropicModel | null;
-export declare function parseOpenAIModel(modelId: string): OpenAIModel | null;
+export declare const parseGeminiModel: (modelId: string) => GeminiModel | null;
+export declare const parseAnthropicModel: (modelId: string) => AnthropicModel | null;
+export declare const parseOpenAIModel: (modelId: string) => OpenAIModel | null;
+/**
+ * Parse a GLM (Zhipu / Z.AI) model id into family + variant + vision + version.
+ * Shape: `glm-<version>[v][-<variant>]` — e.g. `glm-4.5`, `glm-4.5-air`,
+ * `glm-5-turbo`, `glm-4.5v`, `glm-5-preview`. The `v` (vision) attaches to the
+ * version; other variants are `-` suffixes. Standalone like `parseAnthropicModel`
+ * is used in family.ts — GLM needs no global thinking policy, so it stays out of
+ * `parseKnownModel`.
+ */
+export declare const parseGlmModel: (modelId: string) => GlmModel | null;
 export declare function isFableOrMythos(kind: AnthropicKind): boolean;
 export declare function parseSemVer(version: string): SemVer | null;
 export declare function semverGte(left: SemVer | string, right: SemVer | string): boolean;

package/dist/types/identity/family.d.ts

@@ -37,6 +37,28 @@ export declare function isMinimaxM2FamilyModelId(modelId: string): boolean;
  * and `none`.
  */
 export declare function isOpenAIGptOssModelId(modelId: string): boolean;
+/**
+ * Reasoning-capable GLM coding SKUs: glm-4.5 and up on the base / `-air` /
+ * `-turbo` lines. Excludes the vision (`…v`) shape, the non-reasoning
+ * `-flash`/`-flashx`/`-preview` variants, and pre-4.5 ids. Matching the family
+ * keeps newly-bumped integers (`glm-5.3`, `glm-6`, …) covered without a per-id
+ * allowlist.
+ */
+export declare function isReasoningGlmModelId(modelId: string): boolean;
+/** GLM vision SKUs — the `v` that attaches to the version (`glm-4v`, `glm-4.5v`). */
+export declare function isGlmVisionModelId(modelId: string): boolean;
+/**
+ * Coarse vendor-lineage token for "are two models the same family?" checks
+ * (e.g. picking a cross-family reviewer). All Claude point releases share a token,
+ * Claude and GPT differ; namespace prefixes and aggregator mirrors fold onto the
+ * lineage via {@link parseKnownModel}'s `bareModelId` normalization. Opaque and
+ * comparison-only — not a stable key to persist, since the vocabulary tracks new
+ * releases. Returns `""` for ids it cannot classify; callers fall back to the provider.
+ *
+ * Vendor-only by design: a model's kind/variant (opus vs sonnet, codex vs base) is
+ * collapsed onto the single vendor token; use {@link parseKnownModel} for finer breakdowns.
+ */
+export declare function modelFamilyToken(modelId: string): string;
 /**
  * Adaptive thinking `display` is supported starting with Claude Opus 4.7 and
  * the Claude Fable/Mythos 5 generation. Older adaptive-thinking models

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

@@ -22,6 +22,8 @@ export interface ModelManagerOptions<TApi extends Api = Api, TModelsDevPayload =
     staticModels?: readonly ModelSpec<TApi>[];
     /** Optional override for the cache database path. Default: <agent-dir>/models.db. */
     cacheDbPath?: string;
+    /** Optional provider id override for cache namespacing. Defaults to providerId. */
+    cacheProviderId?: string;
     /** Maximum cache age in milliseconds before considered stale. Default: 24h. */
     cacheTtlMs?: number;
     /** When true, a successful dynamic fetch is the complete provider catalog and prunes static-only models. */

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

@@ -25,10 +25,10 @@ export declare const CATALOG_PROVIDERS: readonly [{
     };
 }, {
     readonly id: "amazon-bedrock";
-    readonly defaultModel: "us.anthropic.claude-opus-4-6-v1";
+    readonly defaultModel: "us.anthropic.claude-opus-4-8";
 }, {
     readonly id: "anthropic";
-    readonly defaultModel: "claude-opus-4-6";
+    readonly defaultModel: "claude-opus-4-8";
     readonly createModelManagerOptions: (config: ModelManagerConfig) => import("..").ModelManagerOptions<"anthropic-messages", unknown>;
 }, {
     readonly id: "cerebras";
@@ -136,7 +136,7 @@ export declare const CATALOG_PROVIDERS: readonly [{
     };
 }, {
     readonly id: "litellm";
-    readonly defaultModel: "claude-opus-4-6";
+    readonly defaultModel: "claude-opus-4-8";
     readonly envVars: readonly ["LITELLM_API_KEY"];
     readonly createModelManagerOptions: (config: ModelManagerConfig) => import("..").ModelManagerOptions<"openai-completions", unknown>;
     readonly catalogDiscovery: {
@@ -176,7 +176,7 @@ export declare const CATALOG_PROVIDERS: readonly [{
     };
 }, {
     readonly id: "nanogpt";
-    readonly defaultModel: "openai/gpt-5.4";
+    readonly defaultModel: "openai/gpt-5.5";
     readonly envVars: readonly ["NANO_GPT_API_KEY"];
     readonly createModelManagerOptions: (config: ModelManagerConfig) => import("..").ModelManagerOptions<"openai-completions", unknown>;
     readonly catalogDiscovery: {
@@ -207,12 +207,12 @@ export declare const CATALOG_PROVIDERS: readonly [{
     };
 }, {
     readonly id: "openai";
-    readonly defaultModel: "gpt-5.4";
+    readonly defaultModel: "gpt-5.5";
     readonly envVars: readonly ["OPENAI_API_KEY"];
     readonly createModelManagerOptions: (config: ModelManagerConfig) => import("..").ModelManagerOptions<"openai-responses", unknown>;
 }, {
     readonly id: "openai-codex";
-    readonly defaultModel: "gpt-5.4";
+    readonly defaultModel: "gpt-5.5";
     readonly envVars: readonly ["OPENAI_CODEX_OAUTH_TOKEN"];
     readonly specialModelManager: true;
 }, {
@@ -227,7 +227,7 @@ export declare const CATALOG_PROVIDERS: readonly [{
     readonly createModelManagerOptions: (config: ModelManagerConfig) => import("..").ModelManagerOptions<import("..").Api, unknown>;
 }, {
     readonly id: "openrouter";
-    readonly defaultModel: "openai/gpt-5.4";
+    readonly defaultModel: "openai/gpt-5.5";
     readonly envVars: readonly ["OPENROUTER_API_KEY"];
     readonly createModelManagerOptions: (config: ModelManagerConfig) => import("..").ModelManagerOptions<"openai-completions", unknown>;
     readonly catalogDiscovery: {
@@ -361,7 +361,7 @@ export declare const CATALOG_PROVIDERS: readonly [{
     };
 }, {
     readonly id: "zenmux";
-    readonly defaultModel: "anthropic/claude-opus-4.6";
+    readonly defaultModel: "anthropic/claude-opus-4.8";
     readonly envVars: readonly ["ZENMUX_API_KEY"];
     readonly createModelManagerOptions: (config: ModelManagerConfig) => import("..").ModelManagerOptions<import("..").Api, unknown>;
     readonly catalogDiscovery: {

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

@@ -165,6 +165,7 @@ export declare function isFireworksKimiK2ModelId(modelId: string): boolean;
  * on Fireworks-backed providers, leaving every other model untouched.
  */
 export declare function clampFireworksKimiMaxTokens(modelId: string, candidate: number): number;
+export declare function clampFireworksKimiMaxTokens(modelId: string, candidate: number | null): number | null;
 /**
  * Fireworks DeepSeek V4 accepts effort via `reasoning_effort` but rejects the
  * DeepSeek-native binary `thinking` toggle when both are present.
@@ -343,7 +344,6 @@ export interface AnthropicModelManagerConfig {
     fetch?: FetchImpl;
 }
 export declare function anthropicModelManagerOptions(config?: AnthropicModelManagerConfig): ModelManagerOptions<"anthropic-messages">;
-export { UNK_CONTEXT_WINDOW, UNK_MAX_TOKENS } from "./discovery-constants";
 /** Describes how to map models.dev API data for a single provider. */
 export interface ModelsDevProviderDescriptor {
     /** Key in the models.dev API response JSON (e.g., "anthropic", "amazon-bedrock") */
@@ -386,3 +386,4 @@ export interface ModelsDevProviderDescriptor {
 export declare function mapModelsDevToModels(data: Record<string, unknown>, descriptors: readonly ModelsDevProviderDescriptor[]): ModelSpec<Api>[];
 /** All provider descriptors for models.dev data mapping in generate-models.ts. */
 export declare const MODELS_DEV_PROVIDER_DESCRIPTORS: readonly ModelsDevProviderDescriptor[];
+export {};

package/dist/types/types.d.ts

@@ -161,6 +161,12 @@ export interface OpenAICompat {
     requiresAssistantContentForToolCalls?: boolean;
     /** Whether the provider supports the `tool_choice` parameter. Default: true. */
     supportsToolChoice?: boolean;
+    /**
+     * Whether forced `tool_choice` values (`"required"` or named tools) are accepted.
+     * When false, request builders keep tools available but downgrade forced choices
+     * to provider-default auto selection. Default: true.
+     */
+    supportsForcedToolChoice?: boolean;
     /**
      * Drop reasoning fields (`reasoning_effort`, OpenRouter `reasoning`) for
      * the request when `tool_choice` forces a tool call. Mirrors the Anthropic
@@ -373,8 +379,8 @@ export interface Model<TApi extends Api = Api> {
     };
     /** Premium Copilot requests charged per user-initiated request (defaults to 1). */
     premiumMultiplier?: number;
-    contextWindow: number;
-    maxTokens: number;
+    contextWindow: number | null;
+    maxTokens: number | null;
     /**
      * When `true`, providers MUST omit `max_output_tokens` (Responses) /
      * `max_tokens` / `max_completion_tokens` (Completions) from the outbound

package/dist/types/utils.d.ts

@@ -1,6 +1,9 @@
 export { isRecord } from "@oh-my-pi/pi-utils";
 export declare function toNumber(value: unknown): number | undefined;
 export declare function toPositiveNumber(value: unknown, fallback: number): number;
+export declare function toPositiveNumber(value: unknown, fallback: number | null): number | null;
+/** Positive finite number, or `null` when the value is missing/non-positive. */
+export declare function toPositiveNumberOrNull(value: unknown): number | null;
 export declare function toBoolean(value: unknown): boolean | undefined;
 export declare function isAnthropicOAuthToken(key: string): boolean;
 /**

package/dist/types/wire/gemini-headers.d.ts

@@ -9,6 +9,7 @@ export declare const getGeminiCliHeaders: (modelId?: string) => {
     "Client-Metadata": string;
 };
 export declare const ANTIGRAVITY_SYSTEM_INSTRUCTION: string;
+export declare const ANTIGRAVITY_NO_PREAMBLE_INSTRUCTION = "CRITICAL: NEVER output rule checks, formatting guidelines, constraint checklists (e.g. \"No emdashes\"), or your thinking/personality preambles in the final response. Output only the final response.";
 /**
  * Antigravity / Cloud Code Assist user agent. Lives in its own file so discovery
  * and usage code can read it without pulling the heavy google-gemini-cli provider

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@oh-my-pi/pi-catalog",
-	"version": "15.12.3",
+	"version": "15.13.0",
 	"description": "Model catalog for omp: bundled model database, provider discovery descriptors, model identity, classification, and equivalence",
 	"homepage": "https://omp.sh",
 	"author": "Can Boluk",
@@ -34,11 +34,11 @@
 	},
 	"dependencies": {
 		"@bufbuild/protobuf": "^2.12.0",
-		"@oh-my-pi/pi-utils": "15.12.3",
-		"zod": "4.4.3"
+		"@oh-my-pi/pi-utils": "15.13.0",
+		"zod": "^4"
 	},
 	"devDependencies": {
-		"@oh-my-pi/pi-ai": "15.12.3",
+		"@oh-my-pi/pi-ai": "15.13.0",
 		"@types/bun": "^1.3.14"
 	},
 	"engines": {

package/src/compat/anthropic.ts

@@ -34,11 +34,17 @@ export function buildAnthropicCompat(spec: ModelSpec<"anthropic-messages">): Res
 	const official = isOfficialAnthropicApiUrl(baseUrl);
 	// Z.AI's Anthropic-compatible proxy lives at `api.z.ai/api/anthropic`.
 	const isZai = modelMatchesHost(spec, "zai");
+	// GitHub Copilot's Anthropic-compatible proxy (api.githubcopilot.com/v1/messages)
+	// rejects the per-tool `eager_input_streaming` field with
+	// `tools.0.custom.eager_input_streaming: Extra inputs are not permitted` and
+	// doesn't whitelist the `fine-grained-tool-streaming-2025-05-14` beta either
+	// (issue #2558), so eager tool-input streaming is unavailable on this host.
+	const isCopilot = modelMatchesHost(spec, "githubCopilot");
 	const compat: ResolvedAnthropicCompat = {
 		officialEndpoint: official,
 		disableStrictTools: false,
 		disableAdaptiveThinking: false,
-		supportsEagerToolInputStreaming: true,
+		supportsEagerToolInputStreaming: !isCopilot,
 		// Long cache retention is only sent to the official API by default;
 		// proxies opt in explicitly via `compat.supportsLongCacheRetention: true`.
 		supportsLongCacheRetention: official,

package/src/compat/openai.ts

@@ -217,6 +217,7 @@ export function buildOpenAICompat(spec: ModelSpec<"openai-completions">): Resolv
 		disableReasoningOnForcedToolChoice: isKimiModel || isAnthropicModel,
 		disableReasoningOnToolChoice: isDeepseekFamily && Boolean(spec.reasoning) && !isOpenRouter,
 		supportsToolChoice: !isDirectDeepseekReasoning,
+		supportsForcedToolChoice: true,
 		maxTokensField: useMaxTokens ? "max_tokens" : "max_completion_tokens",
 		requiresToolResultName: isMistral,
 		requiresAssistantAfterToolResult: false,

package/src/discovery/antigravity.ts

@@ -1,4 +1,4 @@
-import * as z from "zod/v4";
+import { z } from "zod/v4";
 import type { ModelSpec } from "../types";
 import { toPositiveNumber } from "../utils";
 import { ANTIGRAVITY_VARIANT_COLLAPSE_TABLE, collapseEffortVariants } from "../variant-collapse";

package/src/discovery/codex.ts

@@ -1,4 +1,4 @@
-import * as z from "zod/v4";
+import { z } from "zod/v4";
 import type { ModelSpec } from "../types";
 import { isRecord } from "../utils";
 import { CODEX_BASE_URL, OPENAI_HEADER_VALUES, OPENAI_HEADERS } from "../wire/codex";

package/src/discovery/cursor.ts

@@ -1,6 +1,6 @@
 import * as http2 from "node:http2";
 import { create, fromBinary, toBinary } from "@bufbuild/protobuf";
-import * as z from "zod/v4";
+import { z } from "zod/v4";
 import { getBundledModels } from "../models";
 import { toModelSpec } from "../provider-models/bundled-references";
 import type { Model, ModelSpec } from "../types";

package/src/discovery/gemini.ts

@@ -1,7 +1,6 @@
-import * as z from "zod/v4";
+import { z } from "zod/v4";
 import { getBundledModels } from "../models";
 import { toModelSpec } from "../provider-models/bundled-references";
-import { UNK_CONTEXT_WINDOW, UNK_MAX_TOKENS } from "../provider-models/discovery-constants";
 import type { FetchImpl, Model, ModelSpec } from "../types";
 
 const GOOGLE_GENERATIVE_AI_BASE_URL = "https://generativelanguage.googleapis.com/v1beta";
@@ -148,7 +147,9 @@ function normalizeBaseUrl(baseUrl?: string): string {
 	return value.replace(/\/+$/, "");
 }
 
-function normalizePositiveInt(value: number | undefined, fallback: number): number {
+function normalizePositiveInt(value: number | undefined, fallback: number): number;
+function normalizePositiveInt(value: number | undefined, fallback: number | null): number | null;
+function normalizePositiveInt(value: number | undefined, fallback: number | null): number | null {
 	if (typeof value !== "number" || !Number.isFinite(value) || value <= 0) {
 		return fallback;
 	}
@@ -178,8 +179,8 @@ function normalizeModel(
 	}
 
 	const reference = bundledById.get(id);
-	const contextWindow = normalizePositiveInt(item.inputTokenLimit, reference?.contextWindow ?? UNK_CONTEXT_WINDOW);
-	const maxTokens = normalizePositiveInt(item.outputTokenLimit, reference?.maxTokens ?? UNK_MAX_TOKENS);
+	const contextWindow = normalizePositiveInt(item.inputTokenLimit, reference?.contextWindow ?? null);
+	const maxTokens = normalizePositiveInt(item.outputTokenLimit, reference?.maxTokens ?? null);
 	const name = normalizeModelName(item.displayName, reference?.name ?? id);
 
 	if (reference) {

package/src/discovery/openai-compatible.ts

@@ -1,5 +1,4 @@
-import * as z from "zod/v4";
-import { UNK_CONTEXT_WINDOW, UNK_MAX_TOKENS } from "../provider-models/discovery-constants";
+import { z } from "zod/v4";
 import type { Api, FetchImpl, ModelSpec, Provider } from "../types";
 
 const MODELS_PATH = "/models";
@@ -165,8 +164,8 @@ export async function fetchOpenAICompatibleModels<TApi extends Api>(
 			reasoning: false,
 			input: ["text"],
 			cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
-			contextWindow: UNK_CONTEXT_WINDOW,
-			maxTokens: UNK_MAX_TOKENS,
+			contextWindow: null,
+			maxTokens: null,
 		};
 
 		// `mapModel` returning null skips the entry (documented contract); only a

package/src/identity/classify.ts

@@ -14,6 +14,7 @@ export type SemVer = {
 export type GeminiKind = "pro" | "flash";
 export type AnthropicKind = "opus" | "sonnet" | "fable" | "mythos";
 export type OpenAIVariant = "base" | "codex" | "codex-max" | "codex-mini" | "codex-spark" | "mini" | "max" | "nano";
+export type GlmVariant = "base" | "air" | "turbo" | "flash" | "flashx" | "preview";
 
 export interface GeminiModel {
 	family: "gemini";
@@ -33,6 +34,15 @@ export interface OpenAIModel {
 	version: SemVer;
 }
 
+export interface GlmModel {
+	family: "glm";
+	/** Suffix variant (`-air`, `-turbo`, `-flash`, `-flashx`, `-preview`); `base` when none. */
+	variant: GlmVariant;
+	/** Vision SKU — the `v` that attaches directly to the version (`glm-4v`, `glm-4.5v`). */
+	vision: boolean;
+	version: SemVer;
+}
+
 export interface UnknownModel {
 	family: "unknown";
 	id: string;
@@ -55,8 +65,26 @@ export function parseKnownModel(modelId: string): ParsedModel {
 	);
 }
 
+/**
+ * Wrap a parse function in a per-id memo cache. Caches the `null` result too, so
+ * repeated misses (the common case — ids of other families) stay O(1) and never
+ * re-run the regex/semver work.
+ */
+function parser<T>(parse: (modelId: string) => T | null): (modelId: string) => T | null {
+	const cache = new Map<string, T | null>();
+	return modelId => {
+		const hit = cache.get(modelId);
+		if (hit !== undefined || cache.has(modelId)) {
+			return hit ?? null;
+		}
+		const result = parse(modelId);
+		cache.set(modelId, result);
+		return result;
+	};
+}
+
 const GEMINI_SUFFIX = "-preview";
-export function parseGeminiModel(modelId: string): GeminiModel | null {
+export const parseGeminiModel = parser((modelId): GeminiModel | null => {
 	if (modelId.endsWith(GEMINI_SUFFIX)) {
 		modelId = modelId.slice(0, -GEMINI_SUFFIX.length);
 	}
@@ -69,9 +97,9 @@ export function parseGeminiModel(modelId: string): GeminiModel | null {
 		return null;
 	}
 	return { family: "gemini", kind: match[2] as GeminiKind, version };
-}
+});
 
-export function parseAnthropicModel(modelId: string): AnthropicModel | null {
+export const parseAnthropicModel = parser((modelId): AnthropicModel | null => {
 	const match = /claude-(opus|sonnet|fable|mythos)-(\d{1,2}(?:[.-]\d{1,2}){0,2})\b/.exec(modelId);
 	if (!match) {
 		return null;
@@ -81,9 +109,9 @@ export function parseAnthropicModel(modelId: string): AnthropicModel | null {
 		return null;
 	}
 	return { family: "anthropic", kind: match[1] as AnthropicKind, version };
-}
+});
 
-export function parseOpenAIModel(modelId: string): OpenAIModel | null {
+export const parseOpenAIModel = parser((modelId): OpenAIModel | null => {
 	const match = /gpt-(\d+(?:\.\d+){0,2})(?:-(codex-spark|codex-mini|codex-max|codex|mini|max|nano))?\b/.exec(modelId);
 	if (!match) {
 		return null;
@@ -93,7 +121,32 @@ export function parseOpenAIModel(modelId: string): OpenAIModel | null {
 		return null;
 	}
 	return { family: "openai", variant: (match[2] as OpenAIVariant | undefined) ?? "base", version };
-}
+});
+
+/**
+ * Parse a GLM (Zhipu / Z.AI) model id into family + variant + vision + version.
+ * Shape: `glm-<version>[v][-<variant>]` — e.g. `glm-4.5`, `glm-4.5-air`,
+ * `glm-5-turbo`, `glm-4.5v`, `glm-5-preview`. The `v` (vision) attaches to the
+ * version; other variants are `-` suffixes. Standalone like `parseAnthropicModel`
+ * is used in family.ts — GLM needs no global thinking policy, so it stays out of
+ * `parseKnownModel`.
+ */
+export const parseGlmModel = parser((modelId): GlmModel | null => {
+	const match = /glm-(\d{1,2}(?:\.\d+)?)(v)?(?:-(air|turbo|flashx|flash|preview))?\b/.exec(modelId);
+	if (!match) {
+		return null;
+	}
+	const version = parseSemVer(match[1]);
+	if (!version) {
+		return null;
+	}
+	return {
+		family: "glm",
+		variant: (match[3] as GlmVariant | undefined) ?? "base",
+		vision: match[2] === "v",
+		version,
+	};
+});
 
 export function isFableOrMythos(kind: AnthropicKind): boolean {
 	return kind === "fable" || kind === "mythos";