@oh-my-pi/pi-catalog 18.1.16 → 18.1.18

package/CHANGELOG.md

@@ -2,6 +2,32 @@
 
 ## [Unreleased]
 
+## [18.1.18] - 2026-09-11
+
+### Added
+
+- `supports-server-compaction` compat axis (`compat.supportsServerCompaction`): whether a model line accepts Anthropic server-side compaction (`compact-2026-01-12`). Class rules enable it for Opus 4.6+, Sonnet 4.6+, and Fable/Mythos 5 on every Anthropic-messages host; the default is `false`.
+
+### Fixed
+
+- OpenCode Go's DeepSeek Flash lanes (`deepseek-flash`, `deepseek-v4.1-flash`) now declare image input. The gateway serves them with vision despite the IDs carrying no vision suffix, so the class-wide `strip-image-input` rule was dropping attachments the endpoint reads; the modality is declared too, since live discovery seeds these lanes text-only ([#11774](https://github.com/can1357/oh-my-pi/pull/11774) by [@STRML](https://github.com/STRML)).
+- Amazon Bedrock OpenAI models, plus unclassified profiles such as opaque application-inference-profile ARNs, now carry the compatibility policy required to preserve image-bearing tool results ([#11681](https://github.com/can1357/oh-my-pi/issues/11681)).
+- DeepSeek V4.1 Flash requests now honor the documented 384K output maximum instead of being capped at 64K ([#11769](https://github.com/can1357/oh-my-pi/issues/11769)).
+
+## [18.1.17] - 2026-09-10
+
+### Added
+
+- Added DeepSeek V4.1 Flash on OpenRouter with image input and low/high/max reasoning levels ([#11592](https://github.com/can1357/oh-my-pi/pull/11592) by [@mazzanfar](https://github.com/mazzanfar)).
+- Added DeepSeek cost estimates that follow published peak/off-peak rates.
+- Added dated, announced price changes to the catalog, so rates switch on their effective date (e.g. DeepSeek Pro moving to Flash rates).
+- Added Command Code as a built-in provider with API-key login, live model discovery, per-model pricing, native OpenAI/Anthropic-compatible routing, cache-aware token usage, and TTFT metrics ([#11391](https://github.com/can1357/oh-my-pi/pull/11391) by [@CherkaSSH](https://github.com/CherkaSSH)).
+
+### Fixed
+
+- Fixed Command Code models outside the verified effort registry offering unsupported reasoning effort controls, and bundled the live Command Code catalog so fresh installs resolve the default model without waiting for discovery ([#11595](https://github.com/can1357/oh-my-pi/pull/11595) by [@H4vC](https://github.com/H4vC)).
+- Fixed the bundled `deepseek-flash` row shipping without context limits: it now carries its documented 1M context / 384K output so offline context accounting enforces the real window.
+
 ## [18.1.16] - 2026-09-09
 
 - Updated Fire Pass (`firepass`) login validation probe to `accounts/fireworks/routers/glm-5p2-fast` and bundled `glm-5.2-fast` and `kimi-k3-fast` models in place of decommissioned `kimi-k2.6-turbo` ([#10859](https://github.com/can1357/oh-my-pi/pull/10859) by [@olegpulatov](https://github.com/olegpulatov)).

package/README.md

@@ -38,6 +38,26 @@ bun add @oh-my-pi/pi-catalog
 
 Ships TypeScript source directly (no build step); requires Bun ≥ 1.3.14.
 
+## Cost calculation
+
+The `models` subpath (also exported from the root) provides timestamp-aware pricing helpers:
+
+| API | Result |
+| --- | --- |
+| `calculateCost(model, usage, timestamp?)` | Updates and returns `usage.cost` using `model.cost`. |
+| `calculateUsageCost(cost, usage, timestamp?)` | Updates and returns `usage.cost` using a `ModelCost`. |
+| `calculateUncachedInputCost(cost, promptInputTokens, timestamp?)` | Returns the cost of a fully uncached prompt. |
+| `getTimeBasedPricingPeriod(cost, timestamp?)` | Returns `"peak"`, `"off-peak"`, or `undefined` without a schedule. |
+| `getNextTimeBasedPricingTransition(cost, timestamp?)` | Returns the next actual peak/off-peak change strictly after the timestamp, or `undefined` if none exists. |
+
+Timestamps are Unix milliseconds; omitted timestamps use the current time for scheduled pricing. Flat token prices are unaffected. Pricing selects the latest applicable effective rate card, then its long-context tier, then the peak/off-peak multiplier. A transition query concerns the recurring tariff, not dated rate-card changes.
+
+`ModelCost.timeBased` is optional typed metadata (`TimeBasedCost`): `offPeakMultiplier`, `peakWindows` (UTC `weekdays`, Sunday = 0, and start-inclusive/end-exclusive `startMinute`/`endMinute`), and optional `effectiveRates`. Each effective rate is a complete `TokenCost` with an `effectiveFrom` Unix-millisecond timestamp and optional `longContext` tier, replacing the base card from that instant.
+
+Pass the request-start timestamp when estimating request usage, then preserve the resulting monetary amounts rather than repricing history at display time. OMP does this using the assistant message timestamp; it is an estimation convention, not a claim about server billing across boundaries. Prefer monetary costs reported by a provider when available.
+
+Schedules are materialized from the [`time-based-cost` KDL axis](src/compat/rules/README.md#time-based-pricing); this does not add a `timeBased` input field to the coding agent's `models.yml`. See [user-facing pricing behavior](../../docs/models.md#usage-costs-and-time-based-pricing) for DeepSeek rates, dates, and footer indicators.
+
 ## References
 
 - [Monorepo README](https://github.com/can1357/oh-my-pi#readme)

package/dist/types/compat/auth-ids.d.ts

@@ -1,4 +1,4 @@
 /** Every provider with an `auth/<id>.kdl` policy. */
-export type AuthProviderId = "abliteration" | "aiand" | "aimlapi" | "alibaba-coding-plan" | "alibaba-token-plan" | "amazon-bedrock" | "anthropic" | "azure" | "baseten" | "bedrock-mantle" | "cerebras" | "cline-pass" | "cloudflare-ai-gateway" | "coreweave" | "cursor" | "deepinfra" | "deepseek" | "devin" | "exa" | "firepass" | "fireworks" | "github-copilot" | "gitlab-duo" | "gitlab-duo-agent" | "gmi-cloud" | "google" | "google-antigravity" | "google-gemini-cli" | "google-vertex" | "groq" | "huggingface" | "kagi" | "kilo" | "kimi-code" | "litellm" | "llama.cpp" | "lm-studio" | "meta" | "minimax" | "minimax-code" | "minimax-code-cn" | "mistral" | "moonshot" | "muse-code" | "nanogpt" | "novita" | "nvidia" | "ollama" | "ollama-cloud" | "openai" | "openai-codex" | "openai-codex-device" | "opencode-go" | "opencode-zen" | "openrouter" | "parallel" | "perplexity" | "qianfan" | "qwen-portal" | "sakana" | "siliconflow" | "siliconflow-cn" | "synthetic" | "tavily" | "together" | "umans" | "venice" | "vercel-ai-gateway" | "vllm" | "wafer-serverless" | "xai" | "xai-oauth" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "yolo-auto" | "zai" | "zai-coding-plan" | "zenmux" | "zhipu-coding-plan";
+export type AuthProviderId = "abliteration" | "aiand" | "aimlapi" | "alibaba-coding-plan" | "alibaba-token-plan" | "amazon-bedrock" | "anthropic" | "azure" | "baseten" | "bedrock-mantle" | "cerebras" | "cline-pass" | "cloudflare-ai-gateway" | "commandcode" | "coreweave" | "cursor" | "deepinfra" | "deepseek" | "devin" | "exa" | "firepass" | "fireworks" | "github-copilot" | "gitlab-duo" | "gitlab-duo-agent" | "gmi-cloud" | "google" | "google-antigravity" | "google-gemini-cli" | "google-vertex" | "groq" | "huggingface" | "kagi" | "kilo" | "kimi-code" | "litellm" | "llama.cpp" | "lm-studio" | "meta" | "minimax" | "minimax-code" | "minimax-code-cn" | "mistral" | "moonshot" | "muse-code" | "nanogpt" | "novita" | "nvidia" | "ollama" | "ollama-cloud" | "openai" | "openai-codex" | "openai-codex-device" | "opencode-go" | "opencode-zen" | "openrouter" | "parallel" | "perplexity" | "qianfan" | "qwen-portal" | "sakana" | "siliconflow" | "siliconflow-cn" | "synthetic" | "tavily" | "together" | "umans" | "venice" | "vercel-ai-gateway" | "vllm" | "wafer-serverless" | "xai" | "xai-oauth" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "yolo-auto" | "zai" | "zai-coding-plan" | "zenmux" | "zhipu-coding-plan";
 /** Providers whose policy declares a `login` flow (the `/login` roster). */
-export type LoginProviderId = "abliteration" | "aiand" | "alibaba-coding-plan" | "alibaba-token-plan" | "anthropic" | "baseten" | "cerebras" | "cline-pass" | "cloudflare-ai-gateway" | "coreweave" | "cursor" | "deepinfra" | "deepseek" | "devin" | "exa" | "firepass" | "fireworks" | "github-copilot" | "gitlab-duo" | "gitlab-duo-agent" | "gmi-cloud" | "google-antigravity" | "google-gemini-cli" | "huggingface" | "kagi" | "kilo" | "kimi-code" | "litellm" | "llama.cpp" | "lm-studio" | "meta" | "minimax-code" | "minimax-code-cn" | "moonshot" | "muse-code" | "nanogpt" | "novita" | "nvidia" | "ollama" | "ollama-cloud" | "openai-codex" | "openai-codex-device" | "opencode-go" | "opencode-zen" | "openrouter" | "parallel" | "perplexity" | "qianfan" | "qwen-portal" | "sakana" | "siliconflow" | "siliconflow-cn" | "synthetic" | "tavily" | "together" | "umans" | "venice" | "vercel-ai-gateway" | "vllm" | "wafer-serverless" | "xai" | "xai-oauth" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "yolo-auto" | "zai" | "zai-coding-plan" | "zenmux" | "zhipu-coding-plan";
+export type LoginProviderId = "abliteration" | "aiand" | "alibaba-coding-plan" | "alibaba-token-plan" | "anthropic" | "baseten" | "cerebras" | "cline-pass" | "cloudflare-ai-gateway" | "commandcode" | "coreweave" | "cursor" | "deepinfra" | "deepseek" | "devin" | "exa" | "firepass" | "fireworks" | "github-copilot" | "gitlab-duo" | "gitlab-duo-agent" | "gmi-cloud" | "google-antigravity" | "google-gemini-cli" | "huggingface" | "kagi" | "kilo" | "kimi-code" | "litellm" | "llama.cpp" | "lm-studio" | "meta" | "minimax-code" | "minimax-code-cn" | "moonshot" | "muse-code" | "nanogpt" | "novita" | "nvidia" | "ollama" | "ollama-cloud" | "openai-codex" | "openai-codex-device" | "opencode-go" | "opencode-zen" | "openrouter" | "parallel" | "perplexity" | "qianfan" | "qwen-portal" | "sakana" | "siliconflow" | "siliconflow-cn" | "synthetic" | "tavily" | "together" | "umans" | "venice" | "vercel-ai-gateway" | "vllm" | "wafer-serverless" | "xai" | "xai-oauth" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "yolo-auto" | "zai" | "zai-coding-plan" | "zenmux" | "zhipu-coding-plan";

package/dist/types/models.d.ts

@@ -4,12 +4,16 @@ export type GeneratedProvider = keyof typeof MODELS;
 export declare function getBundledModel<TApi extends Api = Api>(provider: GeneratedProvider, modelId: string): Model<TApi>;
 export declare function getBundledProviders(): KnownProvider[];
 export declare function getBundledModels(provider: GeneratedProvider): Model<Api>[];
-/** Price a prompt as fully uncached input under its active context-length tier. */
-export declare function calculateUncachedInputCost(cost: ModelCost, promptInputTokens: number): number;
-/** Price one usage record from a token rate card, including active context tiers. */
-export declare function calculateUsageCost(cost: ModelCost, usage: Usage): Usage["cost"];
-/** Price one usage record from its model's token rate card. */
-export declare function calculateCost<TApi extends Api>(model: Model<TApi>, usage: Usage): Usage["cost"];
+/** Return the recurring UTC tariff period, independently of its monetary multiplier. */
+export declare function getTimeBasedPricingPeriod(cost: ModelCost, timestamp?: number): "peak" | "off-peak" | undefined;
+/** Return the next actual peak/off-peak change strictly after the Unix-ms timestamp. */
+export declare function getNextTimeBasedPricingTransition(cost: ModelCost, timestamp?: number): number | undefined;
+/** Price a fully uncached prompt at its request timestamp (Unix ms); scheduled prices default to now. */
+export declare function calculateUncachedInputCost(cost: ModelCost, promptInputTokens: number, timestamp?: number): number;
+/** Price usage at its request timestamp (Unix ms); only scheduled prices default to now. */
+export declare function calculateUsageCost(cost: ModelCost, usage: Usage, timestamp?: number): Usage["cost"];
+/** Price usage at its request timestamp (Unix ms); preserve the resulting monetary amounts for display. */
+export declare function calculateCost<TApi extends Api>(model: Model<TApi>, usage: Usage, timestamp?: number): Usage["cost"];
 /**
  * Check if two models are equal by comparing both their id and provider.
  * Returns false if either model is null or undefined.

package/dist/types/pricing.d.ts

@@ -0,0 +1,5 @@
+import type { TimeBasedCost } from "./types.js";
+/** Validate serialized schedules before admitting cached model rows. */
+export declare function isTimeBasedCost(value: unknown): value is TimeBasedCost;
+/** KDL object children are named records; normalize them once while constructing the model. */
+export declare function materializeTimeBasedCost(value: unknown): TimeBasedCost;

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

@@ -34,6 +34,14 @@ export interface ProviderDescriptor {
     dynamicModelsAuthoritative?: boolean;
     /** Catalog discovery configuration. Only providers with this field participate in generate-models.ts. */
     catalogDiscovery?: CatalogDiscoveryConfig;
+    /**
+     * When true, generator backfills never copy reasoning/input/limits from
+     * same-id rows on other providers into this provider's rows. Set for
+     * providers whose endpoint discovery is the deployment truth and whose
+     * corrections live in KDL, so a stencil.so or canonical-family reference
+     * cannot reintroduce foreign metadata.
+     */
+    skipCrossProviderReferenceFills?: boolean;
 }
 /** A provider descriptor that has catalog discovery configured. */
 export type CatalogProviderDescriptor = ProviderDescriptor & {
@@ -68,6 +76,13 @@ export interface ProviderCatalogEntry {
     readonly dynamicModelsAuthoritative?: boolean;
     /** Catalog discovery configuration for generate-models.ts. */
     readonly catalogDiscovery?: CatalogDiscoveryConfig;
+    /**
+     * When true, generator backfills never copy reasoning/input/limits from
+     * same-id rows on other providers into this provider's rows. Set for
+     * providers whose endpoint discovery is the deployment truth and whose
+     * corrections live in KDL.
+     */
+    readonly skipCrossProviderReferenceFills?: boolean;
     /**
      * Built bespoke by the coding-agent runtime (OAuth-token-driven managers);
      * excluded from `PROVIDER_DESCRIPTORS` even though models are discoverable.

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

@@ -96,6 +96,18 @@ export declare const CATALOG_PROVIDERS: readonly [{
     readonly catalogDiscovery: {
         readonly label: "Cloudflare AI Gateway";
     };
+}, {
+    readonly id: "commandcode";
+    readonly defaultModel: "claude-sonnet-4-6";
+    readonly envVars: readonly ["COMMAND_CODE_API_KEY", "COMMANDCODE_API_KEY"];
+    readonly createModelManagerOptions: (config: ModelManagerConfig) => import("../index.js").ModelManagerOptions<import("../index.js").Api, unknown>;
+    readonly allowUnauthenticated: true;
+    readonly dynamicModelsAuthoritative: true;
+    readonly catalogDiscovery: {
+        readonly label: "Command Code";
+        readonly allowUnauthenticated: true;
+    };
+    readonly skipCrossProviderReferenceFills: true;
 }, {
     readonly id: "cursor";
     readonly defaultModel: "claude-4.6-opus-high";

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

@@ -720,3 +720,28 @@ export declare const MODELS_DEV_CATALOG_PROVIDER_IDS: readonly string[];
  * stricter startup budget and for deterministic timeout tests.
  */
 export declare function modelsDevCatalogFallback(providerId: string, fetchImpl?: FetchImpl, timeoutMs?: number): ModelsDevFallback<Api> | undefined;
+/**
+ * Configuration for the Command Code Provider API model manager.
+ *
+ * `baseUrl` overrides the Provider API base path for testing; it is
+ * normalized to the shared `/provider` root (a trailing `/v1` is stripped)
+ * so Claude ids route to the Anthropic-compatible Messages endpoint at the
+ * root while every other id uses chat completions under `/v1`.
+ */
+export interface CommandCodeModelManagerConfig {
+    apiKey?: string;
+    baseUrl?: string;
+    fetch?: FetchImpl;
+}
+/**
+ * Builds the Command Code model manager: a mixed-protocol OpenAI-compatible
+ * discovery client. The public `/v1/models` catalog is fetched once per
+ * options instance; `mapModel` pins each row's transport from the
+ * `api-routes` table (Claude ids to `anthropic-messages`, everything else to
+ * `openai-completions`) and seeds neutral capability defaults. Reviewed
+ * Command Code policy (effort ladders, pricing, limits, modalities) is
+ * applied later by `buildModel` from `providers/commandcode.kdl` — the
+ * mapper never inherits another provider's reasoning, rates, image support,
+ * or context window.
+ */
+export declare function commandCodeModelManagerOptions(config?: CommandCodeModelManagerConfig): ModelManagerOptions<Api>;

package/dist/types/types.d.ts

@@ -409,6 +409,20 @@ export interface OpenAICompat {
 export interface AnthropicCompat {
     /** Whether thinking requests may include `context_management` and its beta header. Default: true. */
     supportsContextManagement?: boolean;
+    /**
+     * Whether the model lineage supports Anthropic server-side compaction
+     * (`compact-2026-01-12`: the `compact_20260112` edit and replayed
+     * `compaction` blocks). Rule-owned per model line; the beta covers the
+     * adaptive-thinking generation onward and rejects older lines. Default: false.
+     */
+    supportsServerCompaction?: boolean;
+    /**
+     * Whether the model is served by the first-party Anthropic provider (its
+     * default route is the official API). Rule-owned on the provider; the
+     * compaction transport pairs it with a per-request effective-URL check
+     * because reroutes leave it stale-true. Default: false.
+     */
+    firstPartyProvider?: boolean;
     /**
      * Whether requests may carry `output_config.effort` (and its effort beta
      * header). Vertex AI rejects the field/header. Default: true.
@@ -756,6 +770,13 @@ export type ResolvedAnthropicCompat = Required<Omit<AnthropicCompat, "streamIdle
      * env headers, and cache-TTL shaping without per-request URL parsing.
      */
     officialEndpoint: boolean;
+    /**
+     * The model is served by the first-party Anthropic provider. Unlike
+     * `officialEndpoint` (URL-derived per deployment), this is provider
+     * identity for compaction gating, always paired with a per-request
+     * effective-URL check.
+     */
+    firstPartyProvider: boolean;
 };
 /**
  * Compatibility settings for the devin-agent (Codeium Cascade) API. Cascade
@@ -863,9 +884,27 @@ export interface LongContextTokenCost extends TokenCost {
     /** Whether the long-context tier starts exactly at `inputThreshold`. */
     inputThresholdInclusive?: boolean;
 }
-/** Base token rates plus an optional long-context tier. */
+/** Recurring UTC peak interval; weekdays use Sunday = 0, and the end is exclusive. */
+export interface PeakPricingWindow {
+    weekdays: readonly number[];
+    startMinute: number;
+    endMinute: number;
+}
+/** Complete replacement rate card effective from a Unix-millisecond timestamp. */
+export interface EffectiveTokenCost extends TokenCost {
+    effectiveFrom: number;
+    longContext?: LongContextTokenCost;
+}
+/** Scheduled discounts applied after selecting the effective rate card and context tier. */
+export interface TimeBasedCost {
+    offPeakMultiplier: number;
+    peakWindows: readonly PeakPricingWindow[];
+    effectiveRates?: readonly EffectiveTokenCost[];
+}
+/** Base token rates plus optional long-context and time-based pricing. */
 export interface ModelCost extends TokenCost {
     longContext?: LongContextTokenCost;
+    timeBased?: TimeBasedCost;
 }
 /**
  * Exact local content tokenizer family for a model.
@@ -892,6 +931,8 @@ export interface Model<TApi extends Api = Api> {
     requiresGlyphTokenization?: boolean;
     /** Whether this model requires Cursor's tool-schema combiner projection. */
     requiresCursorToolSchemaProjection?: boolean;
+    /** Whether this model requires tool-result images hoisted into sibling user content blocks. */
+    requiresToolResultImageHoisting?: boolean;
     /**
      * Model id to send on the wire when it differs from `id`. Used by catalog
      * variants that present one upstream model under several local entries —
@@ -1065,7 +1106,7 @@ 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" | "supportsComputerUseConfig"> {
+export interface ModelSpec<TApi extends Api = Api> extends Omit<Model<TApi>, "compat" | "identity" | "compatConfig" | "requiresGlyphTokenization" | "requiresCursorToolSchemaProjection" | "requiresToolResultImageHoisting" | "supportsComputerUseConfig"> {
     /** Sparse compatibility overrides; resolved into `Model.compat` by `buildModel`. */
     compat?: CompatConfigOf<TApi>;
 }

package/dist/types/wire/github-copilot.d.ts

@@ -11,6 +11,16 @@ export declare const COPILOT_CAPI_IDENTITY_HEADERS: {
     readonly "Copilot-Harness-Id": "copilot-sdk";
     readonly "Openai-Intent": "conversation-agent";
 };
+/**
+ * Chat-surface identity for Copilot chat and model-policy requests.
+ * Some Business organizations gate premium models per client surface and
+ * block the CLI identity while allowing chat (issue #11372), so chat traffic
+ * defaults to this id with `COPILOT_INTEGRATION_ID` as an explicit pin and a
+ * one-shot CLI retry on denial. Model discovery keeps the CLI identity above:
+ * it unlocks enterprise/experimental models and listing is not policy-gated
+ * the way chat completions are.
+ */
+export declare const COPILOT_CHAT_INTEGRATION_ID: "copilot-chat";
 /**
  * Copilot API version sent on `api.githubcopilot.com` requests (`/models`,
  * chat endpoints). Newer versions unlock tiered context metadata: `/models`
@@ -42,6 +52,15 @@ export declare const COPILOT_DISCOVERY_HEADERS: {
 };
 /** Preserve model-specific headers while enforcing the current Copilot API identity. */
 export declare function mergeCopilotApiHeaders(headers?: Readonly<Record<string, string>>): Record<string, string>;
+/**
+ * Validate an explicit `Copilot-Integration-Id` override.
+ * omp identifies as the Copilot CLI (`copilot-developer-cli`); some Business
+ * organizations allow Chat clients but block CLI/agentic ones, rejecting every
+ * model with HTTP 403 on an otherwise valid token (issue #11372). Callers pass
+ * the raw `COPILOT_INTEGRATION_ID` value; blank or CR/LF-bearing values are
+ * rejected so the header stays well-formed and the default identity applies.
+ */
+export declare function normalizeCopilotIntegrationId(value: unknown): string | undefined;
 export type ParsedGitHubCopilotApiKey = {
     accessToken: string;
     enterpriseUrl?: string;

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@oh-my-pi/pi-catalog",
-	"version": "18.1.16",
+	"version": "18.1.18",
 	"description": "Model catalog for omp: bundled model database, provider discovery descriptors, model identity, classification, and equivalence",
 	"homepage": "https://omp.sh",
 	"author": "Stencil Labs, Inc.",
@@ -35,12 +35,12 @@
 		"gen:proto": "bun scripts/generate-protocols.ts"
 	},
 	"dependencies": {
-		"@oh-my-pi/omptype": "18.1.16",
-		"@oh-my-pi/pi-utils": "18.1.16"
+		"@oh-my-pi/omptype": "18.1.18",
+		"@oh-my-pi/pi-utils": "18.1.18"
 	},
 	"devDependencies": {
 		"@bgotink/kdl": "0.4.0",
-		"@oh-my-pi/pi-ai": "18.1.16",
+		"@oh-my-pi/pi-ai": "18.1.18",
 		"@types/bun": "^1.3.14"
 	},
 	"engines": {

package/src/build.ts

@@ -10,6 +10,7 @@
 import { resolveModelPolicy } from "./compat/resolve";
 import type { ModelIdentity } from "./compat/types";
 import { resolveModelTokenizer } from "./model-tokenizer";
+import { materializeTimeBasedCost } from "./pricing";
 import type { Api, Model, ModelSpec } from "./types";
 import { cleanModelName } from "./utils";
 
@@ -33,8 +34,8 @@ function isInputModalities(value: unknown): value is ("text" | "image")[] {
  * corrections (`cost-patch`, `limits-patch`, `long-context-cost`,
  * `context-window-floor`) overwrite upstream values; selection metadata
  * (`priority`, `apply-patch-tool-type`, `service-tier-cost`,
- * `requires-cursor-tool-schema-projection`) is rule-owned;
- * `context-promotion-target` fills only when the spec left it unset.
+ * `requires-cursor-tool-schema-projection`, `requires-tool-result-image-hoisting`)
+ * is rule-owned; `context-promotion-target` fills only when the spec left it unset.
  */
 function applyCatalogAssignments<TApi extends Api>(model: Model<TApi>, catalog: Record<string, unknown>): void {
 	const serviceTierCost = objectPayload(catalog.serviceTierCost);
@@ -62,6 +63,12 @@ function applyCatalogAssignments<TApi extends Api>(model: Model<TApi>, catalog:
 	} else {
 		delete model.requiresCursorToolSchemaProjection;
 	}
+	const requiresToolResultImageHoisting = catalog.requiresToolResultImageHoisting;
+	if (requiresToolResultImageHoisting === true) {
+		model.requiresToolResultImageHoisting = true;
+	} else {
+		delete model.requiresToolResultImageHoisting;
+	}
 	const contextPromotionTarget = catalog.contextPromotionTarget;
 	if (typeof contextPromotionTarget === "string" && model.contextPromotionTarget === undefined) {
 		model.contextPromotionTarget = contextPromotionTarget;
@@ -126,6 +133,9 @@ export function applyCatalogCorrections(
 		const cacheWrite = numberField(patch, "cacheWrite");
 		if (cacheWrite !== undefined) model.cost.cacheWrite = cacheWrite;
 	}
+	if (catalog.timeBased !== undefined) {
+		model.cost = { ...model.cost, timeBased: materializeTimeBasedCost(catalog.timeBased) };
+	}
 	const limitsPatch = objectPayload(catalog.limitsPatch);
 	if (limitsPatch !== undefined) {
 		const contextWindow = numberField(limitsPatch, "contextWindow");
@@ -208,6 +218,10 @@ export function buildModel<TApi extends Api>(spec: ModelSpec<TApi>): Model<TApi>
 	const supportsComputerUseConfig = explicitComputerUseConfig(spec);
 	const model: Model<TApi> = {
 		...spec,
+		// An exact `thinking-efforts` rule upgrades a stale `reasoning: false`
+		// discovery default (see `resolveThinkingPolicy`); materialize the
+		// correction so transports and the picker see a reasoning-capable model.
+		reasoning: spec.reasoning || policy.thinking !== undefined,
 		name: cleanModelName(spec.name),
 		identity: policy.identity,
 		requiresGlyphTokenization: policy.identity.class === "anthropic",

package/src/compat/auth-ids.ts

@@ -15,6 +15,7 @@ export type AuthProviderId =
 	| "cerebras"
 	| "cline-pass"
 	| "cloudflare-ai-gateway"
+	| "commandcode"
 	| "coreweave"
 	| "cursor"
 	| "deepinfra"
@@ -95,6 +96,7 @@ export type LoginProviderId =
 	| "cerebras"
 	| "cline-pass"
 	| "cloudflare-ai-gateway"
+	| "commandcode"
 	| "coreweave"
 	| "cursor"
 	| "deepinfra"

package/src/compat/axes.ts

@@ -192,6 +192,7 @@ export const AXES: Readonly<Record<string, AxisDef>> = {
 	"disable-adaptive-thinking": wire("disableAdaptiveThinking", ["anthropic"]),
 	"disable-strict-tools": wire("disableStrictTools", ["anthropic"]),
 	"escape-builtin-tool-names": wire("escapeBuiltinToolNames", ["anthropic"]),
+	"first-party-provider": wire("firstPartyProvider", ["anthropic"]),
 	"inject-claude-code-instruction": wire("injectClaudeCodeInstruction", ["anthropic"]),
 	"official-endpoint": wire("officialEndpoint", ["anthropic", "openai-responses"]),
 	"replay-unsigned-thinking": wire("replayUnsignedThinking", ["anthropic"]),
@@ -205,6 +206,7 @@ export const AXES: Readonly<Record<string, AxisDef>> = {
 	"supports-mid-conversation-system": wire("supportsMidConversationSystem", ["anthropic"]),
 	"supports-mid-conversation-tool-changes": wire("supportsMidConversationToolChanges", ["anthropic"]),
 	"supports-per-message-effort": wire("supportsPerMessageEffort", ["anthropic"]),
+	"supports-server-compaction": wire("supportsServerCompaction", ["anthropic"]),
 	"supports-thinking-binding-controls": wire("supportsThinkingBindingControls", ["anthropic"]),
 	"supports-turn-scoped-system": wire("supportsTurnScopedSystem", ["anthropic"]),
 
@@ -259,6 +261,7 @@ export const AXES: Readonly<Record<string, AxisDef>> = {
 	"thinking-prefix-binding": { key: "prefixBinding", set: "thinking", shape: "scalar" },
 	"thinking-suppress-when-off": { key: "suppressWhenOff", set: "thinking", shape: "scalar" },
 	"thinking-supports-display": { key: "supportsDisplay", set: "thinking", shape: "scalar" },
+	"thinking-upgrade-neutral": { key: "upgradeNeutral", set: "thinking", shape: "scalar" },
 
 	// ── catalog metadata ──
 	"apply-patch-tool-type": {
@@ -284,8 +287,14 @@ export const AXES: Readonly<Record<string, AxisDef>> = {
 		set: "catalog",
 		shape: "scalar",
 	},
+	"requires-tool-result-image-hoisting": {
+		key: "requiresToolResultImageHoisting",
+		set: "catalog",
+		shape: "scalar",
+	},
 	priority: { key: "priority", set: "catalog", shape: "scalar" },
 	"service-tier-cost": { key: "serviceTierCost", set: "catalog", shape: "object" },
+	"time-based-cost": { key: "timeBased", set: "catalog", shape: "object" },
 };
 
 /** Records applicable to each API family; used by `resolve.ts` when applying wire axes. */

package/src/compat/resolve.ts

@@ -843,6 +843,8 @@ function resolveAnthropicPolicy(
 		officialEndpoint: official,
 		signingEndpoint,
 		supportsContextManagement: true,
+		supportsServerCompaction: false,
+		firstPartyProvider: false,
 		supportsOutputEffort: true,
 		disableStrictTools: isAzure,
 		disableAdaptiveThinking: false,
@@ -1021,6 +1023,7 @@ interface RuleThinking {
 	suppressWhenOff?: boolean;
 	supportsDisplay?: boolean;
 	prefixBinding?: boolean;
+	upgradeNeutral?: boolean;
 }
 
 function readRuleThinking(axes: ResolvedAxes): RuleThinking {
@@ -1040,6 +1043,7 @@ function readRuleThinking(axes: ResolvedAxes): RuleThinking {
 	if (typeof raw.suppressWhenOff === "boolean") out.suppressWhenOff = raw.suppressWhenOff;
 	if (typeof raw.supportsDisplay === "boolean") out.supportsDisplay = raw.supportsDisplay;
 	if (typeof raw.prefixBinding === "boolean") out.prefixBinding = raw.prefixBinding;
+	if (typeof raw.upgradeNeutral === "boolean") out.upgradeNeutral = raw.upgradeNeutral;
 	return out;
 }
 
@@ -1065,7 +1069,18 @@ function resolveThinkingPolicy<TApi extends Api>(
 	axes: ResolvedAxes,
 	compat: CompatOf<TApi>,
 ): ThinkingConfig | undefined {
-	if (!spec.reasoning) return undefined;
+	const rule = readRuleThinking(axes);
+	const explicitThinking =
+		spec.thinking !== undefined && Array.isArray(spec.thinking.efforts) && spec.thinking.efforts.length > 0
+			? spec.thinking
+			: undefined;
+	// An explicit wire vocabulary is authoritative when discovery reports no
+	// reasoning (e.g. Synthetic's `none`-only off-switch): reviewed KDL must
+	// not re-expand it into an unadvertised ladder. Absent metadata is
+	// repaired only where KDL opts in with `thinking-upgrade-neutral`
+	// alongside an exact `thinking-efforts` ladder (the cascade upgrade for
+	// stale source capability data); otherwise the neutral default holds.
+	if (!spec.reasoning && (explicitThinking !== undefined || rule.upgradeNeutral !== true)) return undefined;
 	if (
 		spec.provider === "cline-pass" &&
 		compat !== undefined &&
@@ -1075,9 +1090,8 @@ function resolveThinkingPolicy<TApi extends Api>(
 		return undefined;
 	}
 	if (omitsWireReasoningEffort(spec.api, compat)) return undefined;
-	const rule = readRuleThinking(axes);
-	if (spec.thinking && Array.isArray(spec.thinking.efforts) && spec.thinking.efforts.length > 0) {
-		return fillExplicitThinking(spec, facts, compat, spec.thinking, rule);
+	if (explicitThinking !== undefined) {
+		return fillExplicitThinking(spec, facts, compat, explicitThinking, rule);
 	}
 	if (compat !== undefined && "trustExplicitThinkingOnly" in compat && compat.trustExplicitThinkingOnly === true) {
 		return undefined;

package/src/compat/rules/README.md

@@ -207,6 +207,43 @@ The three value shapes are:
 A rule cannot assign the same resolved axis twice in one block.
 One object axis carries a computed form: `long-context-cost` accepts either the absolute rates (`input-threshold` + `input`/`output`/`cache-read`/`cache-write`) or `input-threshold` + `multiplier` (with optional `input-threshold-inclusive`), which derives the tier from the row's live base price at build time so the rule tracks upstream list-price updates (xAI's SuperGrok 200K tier). Rows without a token price carry no tier.
 
+### Time-based pricing
+
+The catalog object axis `time-based-cost` materializes into `ModelCost.timeBased`. It uses **named child objects**, not KDL arrays or repeated anonymous windows. For example, inside a matching provider/model scope:
+
+```kdl
+time-based-cost {
+    off-peak-multiplier 0.5
+    peak-windows {
+        morning {
+            weekdays "1,2,3,4,5"
+            start-minute 60
+            end-minute 240
+        }
+        afternoon {
+            weekdays "1,2,3,4,5"
+            start-minute 360
+            end-minute 600
+        }
+    }
+    effective-rates {
+        flash-pricing {
+            effective-from "2026-09-14T04:00:00Z"
+            input 0.30
+            output 1.20
+            cache-read 0.006
+            cache-write 0
+        }
+    }
+}
+```
+
+`morning`, `afternoon`, and `flash-pricing` are arbitrary unique object names, discarded when the payload is normalized into arrays. `weekdays` is a comma-separated string of distinct UTC weekday numbers (`0` = Sunday through `6` = Saturday), without spaces. Each window has integer minutes with `0 <= start-minute < end-minute <= 1440`; its start is inclusive and end exclusive. Split overnight windows across days. Outside the union of peak windows, the nonnegative `off-peak-multiplier` applies to token costs.
+
+`effective-rates` is optional. Each entry requires a distinct valid ISO UTC `effective-from` (`YYYY-MM-DDTHH:mm:ssZ` or with three fractional-second digits) and all four nonnegative per-million-token rates. The latest entry at or before the request timestamp replaces the entire base card; before the first entry, the base card applies. An entry may contain a `long-context` object with `input-threshold`, optional `input-threshold-inclusive`, and all four absolute rates. Effective rates do not inherit the base card's long-context tier. Selection order is effective card, context tier, then tariff multiplier.
+
+The recurring schedule and dated DeepSeek transition above come from [DeepSeek's official pricing](https://api-docs.deepseek.com/quick_start/pricing); see `providers/deepseek.kdl` for the complete rules and [the catalog API](../../../README.md#cost-calculation) for timestamp semantics. This is catalog policy metadata, not a supported schedule syntax for the coding agent's `models.yml`.
+
 ### Precedence and ambiguity
 
 Rules resolve independently per axis. A matching rule is ranked by:

package/src/compat/rules/auth/_order.kdl

@@ -5,7 +5,7 @@ login-order "openai-codex" "anthropic" "zai" "zai-coding-plan" "kimi-code" "open
 	"cursor" "devin" "google-antigravity" "google-gemini-cli" "openai-codex-device" "xai" "xai-oauth" \
 	"gitlab-duo" "gitlab-duo-agent" "alibaba-coding-plan" "alibaba-token-plan" "aiand" "abliteration" "zhipu-coding-plan" \
 	"umans" "qwen-portal" "sakana" "minimax-code" "minimax-code-cn" "xiaomi" "xiaomi-token-plan-sgp" \
-	"xiaomi-token-plan-ams" "xiaomi-token-plan-cn" "firepass" "cline-pass" "deepseek" "muse-code" "meta" "moonshot" \
+	"xiaomi-token-plan-ams" "xiaomi-token-plan-cn" "firepass" "cline-pass" "commandcode" "deepseek" "muse-code" "meta" "moonshot" \
 	"cerebras" "baseten" "fireworks" "together" "nvidia" "novita" "deepinfra" "huggingface" "perplexity" \
 	"qianfan" "venice" "siliconflow" "siliconflow-cn" "synthetic" "nanogpt" "wafer-serverless" "coreweave" \
 	"vercel-ai-gateway" "cloudflare-ai-gateway" "litellm" "kilo" "zenmux" "opencode-zen" "opencode-go" \

package/src/compat/rules/auth/commandcode.kdl

@@ -0,0 +1,13 @@
+// No `validate` probe on purpose: GET /provider/v1/models is public (200
+// with no or bogus credentials, verified 2026-09-09), so a models-endpoint
+// probe could never reject a bad key. A chat-completions probe would bill
+// the key and 403 for Go-plan keys that are still valid CLI credentials.
+auth "commandcode" {
+	name "Command Code"
+	env "COMMAND_CODE_API_KEY" "COMMANDCODE_API_KEY"
+	login "api-key" {
+		auth-url "https://commandcode.ai/studio"
+		instructions "Create or copy a Provider API key from Command Code Studio"
+		prompt "Paste your Command Code API key" placeholder="user_..."
+	}
+}

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

@@ -7,6 +7,30 @@ class "anthropic" {
 			thinking-prefix-binding #true
 		}
 	}
+	// Server-side compaction (compact-2026-01-12) is a lineage capability: the
+	// documented model list is the adaptive-thinking generation onward (Opus
+	// 4.6+, Sonnet 4.6+, Fable/Mythos 5+). Whether a host delivers the beta is
+	// decided at request time from the effective endpoint, not here.
+	family "fable" {
+		revision ">=5" {
+			supports-server-compaction #true
+		}
+	}
+	family "mythos" {
+		revision ">=5" {
+			supports-server-compaction #true
+		}
+	}
+	family "opus" {
+		revision ">=4.6" {
+			supports-server-compaction #true
+		}
+	}
+	family "sonnet" {
+		revision ">=4.6" {
+			supports-server-compaction #true
+		}
+	}
 	on "anthropic" "cloudflare-ai-gateway" "google-vertex" {
 		family "fable" {
 			revision ">=5" {

package/src/compat/rules/providers/amazon-bedrock.kdl

@@ -15,6 +15,8 @@ provider "amazon-bedrock" {
 	}
 	class "openai" {
 		thinking-mode "effort"
+		// OpenAI models reject Converse image blocks nested inside toolResult content.
+		requires-tool-result-image-hoisting #true
 	}
 	class "deepseek" {
 		requires-reasoning-content-for-all-assistant-turns #true
@@ -28,6 +30,11 @@ provider "amazon-bedrock" {
 	class "unknown" {
 		thinking-efforts "minimal" "low" "medium" "high"
 		thinking-mode "budget"
+		// Opaque application-inference-profile ARNs carry no vendor signal, so an
+		// OpenAI-backed profile classifies here and would otherwise nest images in
+		// toolResult content and hit the same rejection. Hoisting is accepted by
+		// every Converse model measured, so it is the safe default for unknowns.
+		requires-tool-result-image-hoisting #true
 	}
 	// residue: taxonomy ranks and exact globs do not isolate these models.
 	models "moonshot.kimi-k2-thinking" {