@xberg-io/liter-llm-wasm 1.9.3 → 1.10.0

package/README.md

@@ -10,7 +10,7 @@
 <div align="center" style="display: flex; flex-wrap: wrap; gap: 8px; justify-content: center; margin: 20px 0">
   <!-- Built with -->
   <a href="https://github.com/xberg-io/alef">
-    <img src="https://img.shields.io/badge/Bindings-alef%20%D7%90-007ec6" alt="Bindings" />
+    <img src="https://img.shields.io/badge/built%20with-alef%20%D7%90-007ec6" alt="Built with alef" />
   </a>
   <!-- Language Bindings -->
   <a href="https://crates.io/crates/liter-llm">
@@ -79,7 +79,7 @@
   </a>
 </div>
 
-Universal LLM API client for browsers and WebAssembly runtimes. Access 143 LLM providers with portable deployment across browsers, Deno, and Cloudflare Workers.
+Universal LLM API client for browsers and WebAssembly runtimes. Access 163 LLM providers with portable deployment across browsers, Deno, and Cloudflare Workers.
 
 ## What This Package Provides
 
@@ -175,7 +175,7 @@ console.log();
 
 ## Features
 
-### Supported Providers (143)
+### Supported Providers (163)
 
 Route to any provider using the `provider/model` prefix convention:
 
@@ -195,7 +195,7 @@ Route to any provider using the `provider/model` prefix convention:
 
 ### Key Capabilities
 
-- **Provider Routing** -- Single client for 143 LLM providers via `provider/model` prefix
+- **Provider Routing** -- Single client for 163 LLM providers via `provider/model` prefix
 - **Local LLMs** — Connect to locally-hosted models via Ollama, LM Studio, vLLM, llama.cpp, and other local inference servers
 - **Unified API** -- Consistent `chat`, `chat_stream`, `embeddings`, `list_models` interface
 - **Streaming** -- Real-time token streaming via `chat_stream`
@@ -216,7 +216,7 @@ Built on a compiled Rust core for speed and safety:
 
 ## Provider Routing
 
-Route to 143 providers using the `provider/model` prefix convention:
+Route to 163 providers using the `provider/model` prefix convention:
 
 ```text
 openai/gpt-4o
@@ -251,7 +251,7 @@ To use the MCP server inside a coding agent, install the **liter-llm plugin** fr
 
 - **[Documentation](https://docs.liter-llm.xberg.io)** -- Full docs and API reference
 - **[GitHub Repository](https://github.com/xberg-io/liter-llm)** -- Source, issues, and discussions
-- **[Provider Registry](https://github.com/xberg-io/liter-llm/blob/main/schemas/providers.json)** -- 143 supported providers
+- **[Provider Registry](https://github.com/xberg-io/liter-llm/blob/main/schemas/providers.json)** -- 163 supported providers
 
 ## Part of Xberg.io
 
@@ -259,7 +259,7 @@ To use the MCP server inside a coding agent, install the **liter-llm plugin** fr
 - [Xberg Enterprise](https://github.com/xberg-io/xberg-enterprise) — managed extraction API with SDKs, dashboards, and observability.
 - [crawlberg](https://github.com/xberg-io/crawlberg) — web crawling and scraping with HTML→Markdown and headless-Chrome fallback.
 - [html-to-markdown](https://github.com/xberg-io/html-to-markdown) — fast, lossless HTML→Markdown engine.
-- [liter-llm](https://github.com/xberg-io/liter-llm) — universal LLM API client with native bindings for 14 languages and 143 providers.
+- [liter-llm](https://github.com/xberg-io/liter-llm) — universal LLM API client with native bindings for 14 languages and 163 providers.
 - [tree-sitter-language-pack](https://github.com/xberg-io/tree-sitter-language-pack) — tree-sitter grammars and code-intelligence primitives.
 - [alef](https://github.com/xberg-io/alef) — the polyglot binding generator that produces every per-language binding across the 5 polyglot repos.
 - [Discord](https://discord.gg/xt9WY3GnKR) — community, roadmap, announcements.

package/liter_llm_wasm.d.ts

@@ -256,6 +256,24 @@ export enum WasmBatchStatus {
     Cancelled = 7,
 }
 
+/**
+ * Plain-data configuration for `refresh_catalog`.
+ *
+ * Deliberately FFI/binding-friendly: no `Duration` or `PathBuf`, just
+ * primitives that translate directly across language boundaries.
+ */
+export class WasmCatalogRefreshConfig {
+    free(): void;
+    [Symbol.dispose](): void;
+    static default(): WasmCatalogRefreshConfig;
+    constructor(enabled?: boolean | null, sourceUrl?: string | null, ttlSeconds?: bigint | null, cachePath?: string | null);
+    get cachePath(): string | undefined;
+    set cachePath(value: string | null | undefined);
+    enabled: boolean;
+    sourceUrl: string;
+    ttlSeconds: bigint;
+}
+
 /**
  * A streamed chunk of a chat completion response.
  */
@@ -541,7 +559,7 @@ export class WasmDecodedDataUrl {
 /**
  * Default client implementation backed by `reqwest`.
  *
- * Sends requests to 143 LLM providers with automatic provider detection
+ * Sends requests to 163 LLM providers with automatic provider detection
  * and per-request routing. The provider is resolved at construction time
  * from `model_hint` (or defaults to OpenAI), but individual requests can
  * override the provider via model name prefix (e.g. `"anthropic/claude-3-5-sonnet"`
@@ -926,6 +944,56 @@ export enum WasmModality {
     Image = 2,
 }
 
+/**
+ * Public, FFI-friendly snapshot of a model's pricing and capability
+ * metadata, projected from `ModelPricing`.
+ *
+ * Unlike `ModelPricing` (which is excluded from binding generation),
+ * `ModelInfo` is an owned plain-data DTO safe to hand across the FFI
+ * boundary — see `model_info`.
+ */
+export class WasmModelInfo {
+    free(): void;
+    [Symbol.dispose](): void;
+    static default(): WasmModelInfo;
+    constructor(inputCostPerToken?: number | null, outputCostPerToken?: number | null, tiers?: WasmModelTier[] | null, cacheReadInputTokenCost?: number | null, cacheCreationInputTokenCost?: number | null, inputCostPerAudioToken?: number | null, outputCostPerAudioToken?: number | null, outputCostPerReasoningToken?: number | null, maxTokens?: bigint | null, maxInputTokens?: bigint | null, maxOutputTokens?: bigint | null, mode?: string | null, supportsVision?: boolean | null, supportsFunctionCalling?: boolean | null, supportsReasoning?: boolean | null, supportsStructuredOutput?: boolean | null, supportsAudioInput?: boolean | null, supportsAudioOutput?: boolean | null, supportsPromptCaching?: boolean | null);
+    get cacheCreationInputTokenCost(): number | undefined;
+    set cacheCreationInputTokenCost(value: number | null | undefined);
+    get cacheReadInputTokenCost(): number | undefined;
+    set cacheReadInputTokenCost(value: number | null | undefined);
+    get inputCostPerAudioToken(): number | undefined;
+    set inputCostPerAudioToken(value: number | null | undefined);
+    inputCostPerToken: number;
+    get maxInputTokens(): bigint | undefined;
+    set maxInputTokens(value: bigint | null | undefined);
+    get maxOutputTokens(): bigint | undefined;
+    set maxOutputTokens(value: bigint | null | undefined);
+    get maxTokens(): bigint | undefined;
+    set maxTokens(value: bigint | null | undefined);
+    get mode(): string | undefined;
+    set mode(value: string | null | undefined);
+    get outputCostPerAudioToken(): number | undefined;
+    set outputCostPerAudioToken(value: number | null | undefined);
+    get outputCostPerReasoningToken(): number | undefined;
+    set outputCostPerReasoningToken(value: number | null | undefined);
+    outputCostPerToken: number;
+    get supportsAudioInput(): boolean | undefined;
+    set supportsAudioInput(value: boolean | null | undefined);
+    get supportsAudioOutput(): boolean | undefined;
+    set supportsAudioOutput(value: boolean | null | undefined);
+    get supportsFunctionCalling(): boolean | undefined;
+    set supportsFunctionCalling(value: boolean | null | undefined);
+    get supportsPromptCaching(): boolean | undefined;
+    set supportsPromptCaching(value: boolean | null | undefined);
+    get supportsReasoning(): boolean | undefined;
+    set supportsReasoning(value: boolean | null | undefined);
+    get supportsStructuredOutput(): boolean | undefined;
+    set supportsStructuredOutput(value: boolean | null | undefined);
+    get supportsVision(): boolean | undefined;
+    set supportsVision(value: boolean | null | undefined);
+    tiers: WasmModelTier[];
+}
+
 /**
  * A model available from the API.
  */
@@ -940,6 +1008,30 @@ export class WasmModelObject {
     ownedBy: string;
 }
 
+/**
+ * Public, FFI-friendly snapshot of a single context-window pricing tier,
+ * projected from `PricingTier`.
+ */
+export class WasmModelTier {
+    free(): void;
+    [Symbol.dispose](): void;
+    static default(): WasmModelTier;
+    constructor(minContextTokens?: bigint | null, inputCostPerToken?: number | null, outputCostPerToken?: number | null, cacheReadInputTokenCost?: number | null, cacheCreationInputTokenCost?: number | null, inputCostPerAudioToken?: number | null, outputCostPerAudioToken?: number | null, outputCostPerReasoningToken?: number | null);
+    get cacheCreationInputTokenCost(): number | undefined;
+    set cacheCreationInputTokenCost(value: number | null | undefined);
+    get cacheReadInputTokenCost(): number | undefined;
+    set cacheReadInputTokenCost(value: number | null | undefined);
+    get inputCostPerAudioToken(): number | undefined;
+    set inputCostPerAudioToken(value: number | null | undefined);
+    inputCostPerToken: number;
+    minContextTokens: bigint;
+    get outputCostPerAudioToken(): number | undefined;
+    set outputCostPerAudioToken(value: number | null | undefined);
+    get outputCostPerReasoningToken(): number | undefined;
+    set outputCostPerReasoningToken(value: number | null | undefined);
+    outputCostPerToken: number;
+}
+
 /**
  * Response listing available models.
  */
@@ -1209,6 +1301,15 @@ export enum WasmReasoningEffort {
     High = 2,
 }
 
+/**
+ * Result of a `refresh_catalog` call.
+ */
+export enum WasmRefreshOutcome {
+    Disabled = 0,
+    FromCache = 1,
+    Fetched = 2,
+}
+
 /**
  * A document to be reranked — either a plain string or an object with a text field.
  */
@@ -1696,7 +1797,7 @@ export function allProviders(): WasmProviderConfig[];
 /**
  * Return the capability flags for a named provider.
  *
- * Performs an O(n) linear scan over the embedded registry (143 entries).
+ * Performs an O(n) linear scan over the embedded registry (163 entries).
  * Returns an owned value so bindings can pass capability data without
  * borrowing registry internals.
  *
@@ -1727,6 +1828,16 @@ export function checkBound(context: string, current_len: number, incoming: numbe
  */
 export function clear(): void;
 
+/**
+ * Clear the overlay registry, reverting `completion_cost`,
+ * `completion_cost_with_cache`, and `model_info` to the
+ * embedded catalog.
+ *
+ * Primarily a test seam (see `install_catalog_overlay_from_str`); also
+ * usable by long-running processes that want to abandon a runtime refresh.
+ */
+export function clearCatalogOverlay(): void;
+
 /**
  * Calculate the estimated cost of a completion given a model name and token
  * counts.
@@ -1755,6 +1866,12 @@ export function completionCost(model: string, prompt_tokens: bigint, completion_
  *
  * Returns `None` if the model is not present in the embedded pricing
  * registry, mirroring `completion_cost`.
+ *
+ * When the model has `ModelPricing.tiers`, the tier whose
+ * `min_context_tokens` is the highest value `<= prompt_tokens` supplies the
+ * input/output/cache rates for the whole call; models without tiers (or
+ * when `prompt_tokens` is below every tier threshold) use the base rates
+ * unchanged, matching the original flat-rate behaviour.
  */
 export function completionCostWithCache(model: string, prompt_tokens: bigint, cached_tokens: bigint, completion_tokens: bigint): number | undefined;
 
@@ -1818,6 +1935,62 @@ export function decodeDataUrl(url: string): WasmDecodedDataUrl | undefined;
  */
 export function encodeDataUrl(bytes: Uint8Array, mime?: string | null): string;
 
+/**
+ * Install the overlay registry from a raw catalog JSON string, bypassing
+ * the network and disk cache entirely.
+ *
+ * Parses and flattens `catalog_json` with the same
+ * `registry_from_catalog_str` logic used for the embedded catalog and the
+ * network refresh path, then atomically swaps it in as the active overlay.
+ * A parse failure returns `CatalogRefreshError.Parse` and leaves any
+ * existing overlay untouched.
+ *
+ * This is primarily a testable seam: it lets tests exercise overlay
+ * installation and the embedded/overlay fallback behavior in
+ * `completion_cost` / `model_info` without a real network
+ * call.
+ */
+export function installCatalogOverlayFromStr(catalog_json: string): void;
+
+/**
+ * Look up FFI-friendly pricing and capability metadata for a model.
+ *
+ * Returns `None` if the model is not present in the active pricing
+ * registry. Uses the same exact-match-then-prefix-fallback resolution as
+ * `model_pricing`; unlike `model_pricing`, the result is an owned
+ * `ModelInfo` value safe to hand across the FFI boundary.
+ *
+ * When a runtime catalog refresh has succeeded, this reflects the refreshed
+ * (overlay) catalog; otherwise it reflects the embedded catalog. See
+ * `model_pricing` for the embedded-only alternative.
+ */
+export function modelInfo(model: string): WasmModelInfo | undefined;
+
+/**
+ * Refresh the runtime catalog overlay per `config`.
+ *
+ * - `config.enabled == false`: returns `Ok(``RefreshOutcome.Disabled``)`
+ *   immediately. No network, filesystem, or overlay activity.
+ * - A fresh on-disk cache (age < `config.ttl_seconds`) exists at the
+ *   resolved cache path (`config.cache_path`, or a default under
+ *   `std.env.temp_dir()`): read + flatten it and install the overlay,
+ *   returning `Ok(``RefreshOutcome.FromCache``)`. No network request is
+ *   made.
+ * - Otherwise: validate `config.source_url` uses `https`
+ *   (`CatalogRefreshError.InsecureUrl` otherwise), fetch it, flatten it,
+ *   install the overlay, best-effort write the raw JSON to the cache path
+ *   (a cache write failure does not fail the refresh), and return
+ *   `Ok(``RefreshOutcome.Fetched``)`.
+ *
+ * On any error return, the overlay is left untouched: the previously
+ * active registry (a prior successful overlay, or the embedded catalog if
+ * none was ever installed) remains in effect. This is what makes the
+ * feature air-gap-safe — an unreachable or invalid `source_url` never
+ * degrades `completion_cost` / `model_info` below embedded-catalog
+ * availability.
+ */
+export function refreshCatalog(config: any): Promise<WasmRefreshOutcome>;
+
 /**
  * Register a custom provider in the global runtime registry.
  *
@@ -1856,6 +2029,7 @@ export interface InitOutput {
     readonly __wbg_wasmbatchlistresponse_free: (a: number, b: number) => void;
     readonly __wbg_wasmbatchobject_free: (a: number, b: number) => void;
     readonly __wbg_wasmbatchrequestcounts_free: (a: number, b: number) => void;
+    readonly __wbg_wasmcatalogrefreshconfig_free: (a: number, b: number) => void;
     readonly __wbg_wasmchatcompletionchunk_free: (a: number, b: number) => void;
     readonly __wbg_wasmchatcompletionrequest_free: (a: number, b: number) => void;
     readonly __wbg_wasmchatcompletionresponse_free: (a: number, b: number) => void;
@@ -1889,8 +2063,10 @@ export interface InitOutput {
     readonly __wbg_wasmjsonschemaformat_free: (a: number, b: number) => void;
     readonly __wbg_wasmliterllmerror_free: (a: number, b: number) => void;
     readonly __wbg_wasmmessage_free: (a: number, b: number) => void;
+    readonly __wbg_wasmmodelinfo_free: (a: number, b: number) => void;
     readonly __wbg_wasmmodelobject_free: (a: number, b: number) => void;
     readonly __wbg_wasmmodelslistresponse_free: (a: number, b: number) => void;
+    readonly __wbg_wasmmodeltier_free: (a: number, b: number) => void;
     readonly __wbg_wasmmoderationcategories_free: (a: number, b: number) => void;
     readonly __wbg_wasmmoderationcategoryscores_free: (a: number, b: number) => void;
     readonly __wbg_wasmmoderationrequest_free: (a: number, b: number) => void;
@@ -1937,6 +2113,7 @@ export interface InitOutput {
     readonly chatstreamiterator_next: (a: number) => any;
     readonly checkBound: (a: number, b: number, c: number, d: number, e: number) => [number, number];
     readonly clear: () => void;
+    readonly clearCatalogOverlay: () => void;
     readonly completionCost: (a: number, b: number, c: bigint, d: bigint) => [number, number];
     readonly completionCostWithCache: (a: number, b: number, c: bigint, d: bigint, e: bigint) => [number, number];
     readonly complexProviderNames: () => [number, number, number, number];
@@ -1944,6 +2121,9 @@ export interface InitOutput {
     readonly createClientFromJson: (a: number, b: number) => [number, number, number];
     readonly decodeDataUrl: (a: number, b: number) => number;
     readonly encodeDataUrl: (a: number, b: number, c: number, d: number) => [number, number];
+    readonly installCatalogOverlayFromStr: (a: number, b: number) => [number, number];
+    readonly modelInfo: (a: number, b: number) => number;
+    readonly refreshCatalog: (a: any) => any;
     readonly registerCustomProvider: (a: number) => [number, number];
     readonly unregisterCustomProvider: (a: number, b: number) => [number, number, number];
     readonly wasmassistantmessage_content: (a: number) => [number, number];
@@ -2041,6 +2221,14 @@ export interface InitOutput {
     readonly wasmbatchrequestcounts_set_failed: (a: number, b: bigint) => void;
     readonly wasmbatchrequestcounts_set_total: (a: number, b: bigint) => void;
     readonly wasmbatchrequestcounts_total: (a: number) => bigint;
+    readonly wasmcatalogrefreshconfig_cachePath: (a: number) => [number, number];
+    readonly wasmcatalogrefreshconfig_default: () => number;
+    readonly wasmcatalogrefreshconfig_enabled: (a: number) => number;
+    readonly wasmcatalogrefreshconfig_new: (a: number, b: number, c: number, d: number, e: bigint, f: number, g: number) => number;
+    readonly wasmcatalogrefreshconfig_set_cachePath: (a: number, b: number, c: number) => void;
+    readonly wasmcatalogrefreshconfig_set_enabled: (a: number, b: number) => void;
+    readonly wasmcatalogrefreshconfig_set_sourceUrl: (a: number, b: number, c: number) => void;
+    readonly wasmcatalogrefreshconfig_sourceUrl: (a: number) => [number, number];
     readonly wasmchatcompletionchunk_choices: (a: number) => [number, number];
     readonly wasmchatcompletionchunk_created: (a: number) => bigint;
     readonly wasmchatcompletionchunk_default: () => number;
@@ -2375,6 +2563,36 @@ export interface InitOutput {
     readonly wasmmessage_role: (a: number) => [number, number];
     readonly wasmmessage_set_0: (a: number, b: number) => void;
     readonly wasmmessage_set_role: (a: number, b: number, c: number) => void;
+    readonly wasmmodelinfo_default: () => number;
+    readonly wasmmodelinfo_inputCostPerAudioToken: (a: number) => [number, number];
+    readonly wasmmodelinfo_inputCostPerToken: (a: number) => number;
+    readonly wasmmodelinfo_maxInputTokens: (a: number) => [number, bigint];
+    readonly wasmmodelinfo_maxOutputTokens: (a: number) => [number, bigint];
+    readonly wasmmodelinfo_mode: (a: number) => [number, number];
+    readonly wasmmodelinfo_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: bigint, s: number, t: bigint, u: number, v: bigint, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number) => number;
+    readonly wasmmodelinfo_outputCostPerToken: (a: number) => number;
+    readonly wasmmodelinfo_set_inputCostPerAudioToken: (a: number, b: number, c: number) => void;
+    readonly wasmmodelinfo_set_inputCostPerToken: (a: number, b: number) => void;
+    readonly wasmmodelinfo_set_maxInputTokens: (a: number, b: number, c: bigint) => void;
+    readonly wasmmodelinfo_set_maxOutputTokens: (a: number, b: number, c: bigint) => void;
+    readonly wasmmodelinfo_set_mode: (a: number, b: number, c: number) => void;
+    readonly wasmmodelinfo_set_outputCostPerToken: (a: number, b: number) => void;
+    readonly wasmmodelinfo_set_supportsAudioInput: (a: number, b: number) => void;
+    readonly wasmmodelinfo_set_supportsAudioOutput: (a: number, b: number) => void;
+    readonly wasmmodelinfo_set_supportsFunctionCalling: (a: number, b: number) => void;
+    readonly wasmmodelinfo_set_supportsPromptCaching: (a: number, b: number) => void;
+    readonly wasmmodelinfo_set_supportsReasoning: (a: number, b: number) => void;
+    readonly wasmmodelinfo_set_supportsStructuredOutput: (a: number, b: number) => void;
+    readonly wasmmodelinfo_set_supportsVision: (a: number, b: number) => void;
+    readonly wasmmodelinfo_set_tiers: (a: number, b: number, c: number) => void;
+    readonly wasmmodelinfo_supportsAudioInput: (a: number) => number;
+    readonly wasmmodelinfo_supportsAudioOutput: (a: number) => number;
+    readonly wasmmodelinfo_supportsFunctionCalling: (a: number) => number;
+    readonly wasmmodelinfo_supportsPromptCaching: (a: number) => number;
+    readonly wasmmodelinfo_supportsReasoning: (a: number) => number;
+    readonly wasmmodelinfo_supportsStructuredOutput: (a: number) => number;
+    readonly wasmmodelinfo_supportsVision: (a: number) => number;
+    readonly wasmmodelinfo_tiers: (a: number) => [number, number];
     readonly wasmmodelobject_default: () => number;
     readonly wasmmodelobject_id: (a: number) => [number, number];
     readonly wasmmodelobject_new: (a: number, b: number, c: number, d: number, e: number, f: bigint, g: number, h: number) => number;
@@ -2389,6 +2607,14 @@ export interface InitOutput {
     readonly wasmmodelslistresponse_object: (a: number) => [number, number];
     readonly wasmmodelslistresponse_set_data: (a: number, b: number, c: number) => void;
     readonly wasmmodelslistresponse_set_object: (a: number, b: number, c: number) => void;
+    readonly wasmmodeltier_default: () => number;
+    readonly wasmmodeltier_inputCostPerToken: (a: number) => number;
+    readonly wasmmodeltier_minContextTokens: (a: number) => bigint;
+    readonly wasmmodeltier_new: (a: number, b: bigint, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number) => number;
+    readonly wasmmodeltier_outputCostPerToken: (a: number) => number;
+    readonly wasmmodeltier_set_inputCostPerToken: (a: number, b: number) => void;
+    readonly wasmmodeltier_set_minContextTokens: (a: number, b: bigint) => void;
+    readonly wasmmodeltier_set_outputCostPerToken: (a: number, b: number) => void;
     readonly wasmmoderationcategories_default: () => number;
     readonly wasmmoderationcategories_harassment: (a: number) => number;
     readonly wasmmoderationcategories_harassmentThreatening: (a: number) => number;
@@ -2702,15 +2928,34 @@ export interface InitOutput {
     readonly wasmocrrequest_includeImageBase64: (a: number) => number;
     readonly wasmchatcompletionrequest_set_maxTokens: (a: number, b: number, c: bigint) => void;
     readonly wasmcreateresponserequest_set_maxOutputTokens: (a: number, b: number, c: bigint) => void;
+    readonly wasmmodelinfo_set_cacheCreationInputTokenCost: (a: number, b: number, c: number) => void;
+    readonly wasmmodelinfo_set_maxTokens: (a: number, b: number, c: bigint) => void;
+    readonly wasmmodelinfo_set_outputCostPerAudioToken: (a: number, b: number, c: number) => void;
+    readonly wasmmodelinfo_set_outputCostPerReasoningToken: (a: number, b: number, c: number) => void;
+    readonly wasmmodeltier_set_cacheCreationInputTokenCost: (a: number, b: number, c: number) => void;
+    readonly wasmmodeltier_set_inputCostPerAudioToken: (a: number, b: number, c: number) => void;
+    readonly wasmmodeltier_set_outputCostPerAudioToken: (a: number, b: number, c: number) => void;
+    readonly wasmmodeltier_set_outputCostPerReasoningToken: (a: number, b: number, c: number) => void;
     readonly wasmchatcompletionrequest_maxTokens: (a: number) => [number, bigint];
     readonly wasmcreateresponserequest_maxOutputTokens: (a: number) => [number, bigint];
     readonly wasmcreateresponserequest_temperature: (a: number) => [number, number];
     readonly wasmcreatespeechrequest_speed: (a: number) => [number, number];
     readonly wasmcreatetranscriptionrequest_temperature: (a: number) => [number, number];
+    readonly wasmmodelinfo_cacheCreationInputTokenCost: (a: number) => [number, number];
+    readonly wasmmodelinfo_cacheReadInputTokenCost: (a: number) => [number, number];
+    readonly wasmmodelinfo_maxTokens: (a: number) => [number, bigint];
+    readonly wasmmodelinfo_outputCostPerAudioToken: (a: number) => [number, number];
+    readonly wasmmodelinfo_outputCostPerReasoningToken: (a: number) => [number, number];
+    readonly wasmmodeltier_cacheCreationInputTokenCost: (a: number) => [number, number];
+    readonly wasmmodeltier_cacheReadInputTokenCost: (a: number) => [number, number];
+    readonly wasmmodeltier_inputCostPerAudioToken: (a: number) => [number, number];
+    readonly wasmmodeltier_outputCostPerAudioToken: (a: number) => [number, number];
+    readonly wasmmodeltier_outputCostPerReasoningToken: (a: number) => [number, number];
     readonly wasmtranscriptionresponse_duration: (a: number) => [number, number];
     readonly wasmwaitforbatchconfig_timeoutSecs: (a: number) => [number, number];
     readonly wasmresponseusage_default: () => number;
     readonly wasmprovidercapabilities_set_vision: (a: number, b: number) => void;
+    readonly wasmcatalogrefreshconfig_set_ttlSeconds: (a: number, b: bigint) => void;
     readonly wasmchatcompletionresponse_set_created: (a: number, b: bigint) => void;
     readonly wasmfileobject_set_bytes: (a: number, b: bigint) => void;
     readonly wasmfileobject_set_createdAt: (a: number, b: bigint) => void;
@@ -2735,6 +2980,7 @@ export interface InitOutput {
     readonly wasmfilelistquery_set_limit: (a: number, b: number) => void;
     readonly wasmrerankrequest_set_topN: (a: number, b: number) => void;
     readonly wasmsearchrequest_set_maxResults: (a: number, b: number) => void;
+    readonly wasmcatalogrefreshconfig_ttlSeconds: (a: number) => bigint;
     readonly wasmchatcompletionresponse_created: (a: number) => bigint;
     readonly wasmfileobject_bytes: (a: number) => bigint;
     readonly wasmfileobject_createdAt: (a: number) => bigint;
@@ -2781,7 +3027,6 @@ export interface InitOutput {
     readonly wasmusermessage_default: () => number;
     readonly wasmsystemmessage_set_content: (a: number, b: number) => void;
     readonly wasmusermessage_set_content: (a: number, b: number) => void;
-    readonly wasmresponsetool_default: () => number;
     readonly wasmocrrequest_set_includeImageBase64: (a: number, b: number) => void;
     readonly wasmassistantpart_new: () => number;
     readonly wasmcontentpart_new: () => number;
@@ -2804,8 +3049,11 @@ export interface InitOutput {
     readonly wasmcreateresponserequest_set_temperature: (a: number, b: number, c: number) => void;
     readonly wasmcreatespeechrequest_set_speed: (a: number, b: number, c: number) => void;
     readonly wasmcreatetranscriptionrequest_set_temperature: (a: number, b: number, c: number) => void;
+    readonly wasmmodelinfo_set_cacheReadInputTokenCost: (a: number, b: number, c: number) => void;
+    readonly wasmmodeltier_set_cacheReadInputTokenCost: (a: number, b: number, c: number) => void;
     readonly wasmtranscriptionresponse_set_duration: (a: number, b: number, c: number) => void;
     readonly wasmwaitforbatchconfig_set_timeoutSecs: (a: number, b: number, c: number) => void;
+    readonly wasmresponsetool_default: () => number;
     readonly __wbg_intounderlyingsource_free: (a: number, b: number) => void;
     readonly intounderlyingsource_cancel: (a: number) => void;
     readonly intounderlyingsource_pull: (a: number, b: any) => any;

package/liter_llm_wasm.js

@@ -1335,6 +1335,124 @@ export const WasmBatchStatus = Object.freeze({
     Cancelled: 7, "7": "Cancelled",
 });
 
+/**
+ * Plain-data configuration for `refresh_catalog`.
+ *
+ * Deliberately FFI/binding-friendly: no `Duration` or `PathBuf`, just
+ * primitives that translate directly across language boundaries.
+ */
+export class WasmCatalogRefreshConfig {
+    static __wrap(ptr) {
+        const obj = Object.create(WasmCatalogRefreshConfig.prototype);
+        obj.__wbg_ptr = ptr;
+        WasmCatalogRefreshConfigFinalization.register(obj, obj.__wbg_ptr, obj);
+        return obj;
+    }
+    __destroy_into_raw() {
+        const ptr = this.__wbg_ptr;
+        this.__wbg_ptr = 0;
+        WasmCatalogRefreshConfigFinalization.unregister(this);
+        return ptr;
+    }
+    free() {
+        const ptr = this.__destroy_into_raw();
+        wasm.__wbg_wasmcatalogrefreshconfig_free(ptr, 0);
+    }
+    /**
+     * @returns {string | undefined}
+     */
+    get cachePath() {
+        const ret = wasm.wasmcatalogrefreshconfig_cachePath(this.__wbg_ptr);
+        let v1;
+        if (ret[0] !== 0) {
+            v1 = getStringFromWasm0(ret[0], ret[1]).slice();
+            wasm.__wbindgen_free(ret[0], ret[1] * 1, 1);
+        }
+        return v1;
+    }
+    /**
+     * @returns {WasmCatalogRefreshConfig}
+     */
+    static default() {
+        const ret = wasm.wasmcatalogrefreshconfig_default();
+        return WasmCatalogRefreshConfig.__wrap(ret);
+    }
+    /**
+     * @returns {boolean}
+     */
+    get enabled() {
+        const ret = wasm.wasmcatalogrefreshconfig_enabled(this.__wbg_ptr);
+        return ret !== 0;
+    }
+    /**
+     * @param {boolean | null} [enabled]
+     * @param {string | null} [sourceUrl]
+     * @param {bigint | null} [ttlSeconds]
+     * @param {string | null} [cachePath]
+     */
+    constructor(enabled, sourceUrl, ttlSeconds, cachePath) {
+        var ptr0 = isLikeNone(sourceUrl) ? 0 : passStringToWasm0(sourceUrl, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        var len0 = WASM_VECTOR_LEN;
+        var ptr1 = isLikeNone(cachePath) ? 0 : passStringToWasm0(cachePath, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        var len1 = WASM_VECTOR_LEN;
+        const ret = wasm.wasmcatalogrefreshconfig_new(isLikeNone(enabled) ? 0xFFFFFF : enabled ? 1 : 0, ptr0, len0, !isLikeNone(ttlSeconds), isLikeNone(ttlSeconds) ? BigInt(0) : ttlSeconds, ptr1, len1);
+        this.__wbg_ptr = ret;
+        WasmCatalogRefreshConfigFinalization.register(this, this.__wbg_ptr, this);
+        return this;
+    }
+    /**
+     * @param {string | null} [value]
+     */
+    set cachePath(value) {
+        var ptr0 = isLikeNone(value) ? 0 : passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        var len0 = WASM_VECTOR_LEN;
+        wasm.wasmcatalogrefreshconfig_set_cachePath(this.__wbg_ptr, ptr0, len0);
+    }
+    /**
+     * @param {boolean} value
+     */
+    set enabled(value) {
+        wasm.wasmcatalogrefreshconfig_set_enabled(this.__wbg_ptr, value);
+    }
+    /**
+     * @param {string} value
+     */
+    set sourceUrl(value) {
+        const ptr0 = passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        const len0 = WASM_VECTOR_LEN;
+        wasm.wasmcatalogrefreshconfig_set_sourceUrl(this.__wbg_ptr, ptr0, len0);
+    }
+    /**
+     * @param {bigint} value
+     */
+    set ttlSeconds(value) {
+        wasm.wasmcatalogrefreshconfig_set_ttlSeconds(this.__wbg_ptr, value);
+    }
+    /**
+     * @returns {string}
+     */
+    get sourceUrl() {
+        let deferred1_0;
+        let deferred1_1;
+        try {
+            const ret = wasm.wasmcatalogrefreshconfig_sourceUrl(this.__wbg_ptr);
+            deferred1_0 = ret[0];
+            deferred1_1 = ret[1];
+            return getStringFromWasm0(ret[0], ret[1]);
+        } finally {
+            wasm.__wbindgen_free(deferred1_0, deferred1_1, 1);
+        }
+    }
+    /**
+     * @returns {bigint}
+     */
+    get ttlSeconds() {
+        const ret = wasm.wasmcatalogrefreshconfig_ttlSeconds(this.__wbg_ptr);
+        return BigInt.asUintN(64, ret);
+    }
+}
+if (Symbol.dispose) WasmCatalogRefreshConfig.prototype[Symbol.dispose] = WasmCatalogRefreshConfig.prototype.free;
+
 /**
  * A streamed chunk of a chat completion response.
  */
@@ -3645,7 +3763,7 @@ if (Symbol.dispose) WasmDecodedDataUrl.prototype[Symbol.dispose] = WasmDecodedDa
 /**
  * Default client implementation backed by `reqwest`.
  *
- * Sends requests to 143 LLM providers with automatic provider detection
+ * Sends requests to 163 LLM providers with automatic provider detection
  * and per-request routing. The provider is resolved at construction time
  * from `model_hint` (or defaults to OpenAI), but individual requests can
  * override the provider via model name prefix (e.g. `"anthropic/claude-3-5-sonnet"`
@@ -5869,6 +5987,330 @@ export const WasmModality = Object.freeze({
     Image: 2, "2": "Image",
 });
 
+/**
+ * Public, FFI-friendly snapshot of a model's pricing and capability
+ * metadata, projected from `ModelPricing`.
+ *
+ * Unlike `ModelPricing` (which is excluded from binding generation),
+ * `ModelInfo` is an owned plain-data DTO safe to hand across the FFI
+ * boundary — see `model_info`.
+ */
+export class WasmModelInfo {
+    static __wrap(ptr) {
+        const obj = Object.create(WasmModelInfo.prototype);
+        obj.__wbg_ptr = ptr;
+        WasmModelInfoFinalization.register(obj, obj.__wbg_ptr, obj);
+        return obj;
+    }
+    __destroy_into_raw() {
+        const ptr = this.__wbg_ptr;
+        this.__wbg_ptr = 0;
+        WasmModelInfoFinalization.unregister(this);
+        return ptr;
+    }
+    free() {
+        const ptr = this.__destroy_into_raw();
+        wasm.__wbg_wasmmodelinfo_free(ptr, 0);
+    }
+    /**
+     * @returns {number | undefined}
+     */
+    get cacheCreationInputTokenCost() {
+        const ret = wasm.wasmmodelinfo_cacheCreationInputTokenCost(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : ret[1];
+    }
+    /**
+     * @returns {number | undefined}
+     */
+    get cacheReadInputTokenCost() {
+        const ret = wasm.wasmmodelinfo_cacheReadInputTokenCost(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : ret[1];
+    }
+    /**
+     * @returns {WasmModelInfo}
+     */
+    static default() {
+        const ret = wasm.wasmmodelinfo_default();
+        return WasmModelInfo.__wrap(ret);
+    }
+    /**
+     * @returns {number | undefined}
+     */
+    get inputCostPerAudioToken() {
+        const ret = wasm.wasmmodelinfo_inputCostPerAudioToken(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : ret[1];
+    }
+    /**
+     * @returns {number}
+     */
+    get inputCostPerToken() {
+        const ret = wasm.wasmmodelinfo_inputCostPerToken(this.__wbg_ptr);
+        return ret;
+    }
+    /**
+     * @returns {bigint | undefined}
+     */
+    get maxInputTokens() {
+        const ret = wasm.wasmmodelinfo_maxInputTokens(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : BigInt.asUintN(64, ret[1]);
+    }
+    /**
+     * @returns {bigint | undefined}
+     */
+    get maxOutputTokens() {
+        const ret = wasm.wasmmodelinfo_maxOutputTokens(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : BigInt.asUintN(64, ret[1]);
+    }
+    /**
+     * @returns {bigint | undefined}
+     */
+    get maxTokens() {
+        const ret = wasm.wasmmodelinfo_maxTokens(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : BigInt.asUintN(64, ret[1]);
+    }
+    /**
+     * @returns {string | undefined}
+     */
+    get mode() {
+        const ret = wasm.wasmmodelinfo_mode(this.__wbg_ptr);
+        let v1;
+        if (ret[0] !== 0) {
+            v1 = getStringFromWasm0(ret[0], ret[1]).slice();
+            wasm.__wbindgen_free(ret[0], ret[1] * 1, 1);
+        }
+        return v1;
+    }
+    /**
+     * @param {number | null} [inputCostPerToken]
+     * @param {number | null} [outputCostPerToken]
+     * @param {WasmModelTier[] | null} [tiers]
+     * @param {number | null} [cacheReadInputTokenCost]
+     * @param {number | null} [cacheCreationInputTokenCost]
+     * @param {number | null} [inputCostPerAudioToken]
+     * @param {number | null} [outputCostPerAudioToken]
+     * @param {number | null} [outputCostPerReasoningToken]
+     * @param {bigint | null} [maxTokens]
+     * @param {bigint | null} [maxInputTokens]
+     * @param {bigint | null} [maxOutputTokens]
+     * @param {string | null} [mode]
+     * @param {boolean | null} [supportsVision]
+     * @param {boolean | null} [supportsFunctionCalling]
+     * @param {boolean | null} [supportsReasoning]
+     * @param {boolean | null} [supportsStructuredOutput]
+     * @param {boolean | null} [supportsAudioInput]
+     * @param {boolean | null} [supportsAudioOutput]
+     * @param {boolean | null} [supportsPromptCaching]
+     */
+    constructor(inputCostPerToken, outputCostPerToken, tiers, cacheReadInputTokenCost, cacheCreationInputTokenCost, inputCostPerAudioToken, outputCostPerAudioToken, outputCostPerReasoningToken, maxTokens, maxInputTokens, maxOutputTokens, mode, supportsVision, supportsFunctionCalling, supportsReasoning, supportsStructuredOutput, supportsAudioInput, supportsAudioOutput, supportsPromptCaching) {
+        var ptr0 = isLikeNone(tiers) ? 0 : passArrayJsValueToWasm0(tiers, wasm.__wbindgen_malloc);
+        var len0 = WASM_VECTOR_LEN;
+        var ptr1 = isLikeNone(mode) ? 0 : passStringToWasm0(mode, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        var len1 = WASM_VECTOR_LEN;
+        const ret = wasm.wasmmodelinfo_new(!isLikeNone(inputCostPerToken), isLikeNone(inputCostPerToken) ? 0 : inputCostPerToken, !isLikeNone(outputCostPerToken), isLikeNone(outputCostPerToken) ? 0 : outputCostPerToken, ptr0, len0, !isLikeNone(cacheReadInputTokenCost), isLikeNone(cacheReadInputTokenCost) ? 0 : cacheReadInputTokenCost, !isLikeNone(cacheCreationInputTokenCost), isLikeNone(cacheCreationInputTokenCost) ? 0 : cacheCreationInputTokenCost, !isLikeNone(inputCostPerAudioToken), isLikeNone(inputCostPerAudioToken) ? 0 : inputCostPerAudioToken, !isLikeNone(outputCostPerAudioToken), isLikeNone(outputCostPerAudioToken) ? 0 : outputCostPerAudioToken, !isLikeNone(outputCostPerReasoningToken), isLikeNone(outputCostPerReasoningToken) ? 0 : outputCostPerReasoningToken, !isLikeNone(maxTokens), isLikeNone(maxTokens) ? BigInt(0) : maxTokens, !isLikeNone(maxInputTokens), isLikeNone(maxInputTokens) ? BigInt(0) : maxInputTokens, !isLikeNone(maxOutputTokens), isLikeNone(maxOutputTokens) ? BigInt(0) : maxOutputTokens, ptr1, len1, isLikeNone(supportsVision) ? 0xFFFFFF : supportsVision ? 1 : 0, isLikeNone(supportsFunctionCalling) ? 0xFFFFFF : supportsFunctionCalling ? 1 : 0, isLikeNone(supportsReasoning) ? 0xFFFFFF : supportsReasoning ? 1 : 0, isLikeNone(supportsStructuredOutput) ? 0xFFFFFF : supportsStructuredOutput ? 1 : 0, isLikeNone(supportsAudioInput) ? 0xFFFFFF : supportsAudioInput ? 1 : 0, isLikeNone(supportsAudioOutput) ? 0xFFFFFF : supportsAudioOutput ? 1 : 0, isLikeNone(supportsPromptCaching) ? 0xFFFFFF : supportsPromptCaching ? 1 : 0);
+        this.__wbg_ptr = ret;
+        WasmModelInfoFinalization.register(this, this.__wbg_ptr, this);
+        return this;
+    }
+    /**
+     * @returns {number | undefined}
+     */
+    get outputCostPerAudioToken() {
+        const ret = wasm.wasmmodelinfo_outputCostPerAudioToken(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : ret[1];
+    }
+    /**
+     * @returns {number | undefined}
+     */
+    get outputCostPerReasoningToken() {
+        const ret = wasm.wasmmodelinfo_outputCostPerReasoningToken(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : ret[1];
+    }
+    /**
+     * @returns {number}
+     */
+    get outputCostPerToken() {
+        const ret = wasm.wasmmodelinfo_outputCostPerToken(this.__wbg_ptr);
+        return ret;
+    }
+    /**
+     * @param {number | null} [value]
+     */
+    set cacheCreationInputTokenCost(value) {
+        wasm.wasmmodelinfo_set_cacheCreationInputTokenCost(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? 0 : value);
+    }
+    /**
+     * @param {number | null} [value]
+     */
+    set cacheReadInputTokenCost(value) {
+        wasm.wasmmodelinfo_set_cacheReadInputTokenCost(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? 0 : value);
+    }
+    /**
+     * @param {number | null} [value]
+     */
+    set inputCostPerAudioToken(value) {
+        wasm.wasmmodelinfo_set_inputCostPerAudioToken(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? 0 : value);
+    }
+    /**
+     * @param {number} value
+     */
+    set inputCostPerToken(value) {
+        wasm.wasmmodelinfo_set_inputCostPerToken(this.__wbg_ptr, value);
+    }
+    /**
+     * @param {bigint | null} [value]
+     */
+    set maxInputTokens(value) {
+        wasm.wasmmodelinfo_set_maxInputTokens(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? BigInt(0) : value);
+    }
+    /**
+     * @param {bigint | null} [value]
+     */
+    set maxOutputTokens(value) {
+        wasm.wasmmodelinfo_set_maxOutputTokens(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? BigInt(0) : value);
+    }
+    /**
+     * @param {bigint | null} [value]
+     */
+    set maxTokens(value) {
+        wasm.wasmmodelinfo_set_maxTokens(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? BigInt(0) : value);
+    }
+    /**
+     * @param {string | null} [value]
+     */
+    set mode(value) {
+        var ptr0 = isLikeNone(value) ? 0 : passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        var len0 = WASM_VECTOR_LEN;
+        wasm.wasmmodelinfo_set_mode(this.__wbg_ptr, ptr0, len0);
+    }
+    /**
+     * @param {number | null} [value]
+     */
+    set outputCostPerAudioToken(value) {
+        wasm.wasmmodelinfo_set_outputCostPerAudioToken(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? 0 : value);
+    }
+    /**
+     * @param {number | null} [value]
+     */
+    set outputCostPerReasoningToken(value) {
+        wasm.wasmmodelinfo_set_outputCostPerReasoningToken(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? 0 : value);
+    }
+    /**
+     * @param {number} value
+     */
+    set outputCostPerToken(value) {
+        wasm.wasmmodelinfo_set_outputCostPerToken(this.__wbg_ptr, value);
+    }
+    /**
+     * @param {boolean | null} [value]
+     */
+    set supportsAudioInput(value) {
+        wasm.wasmmodelinfo_set_supportsAudioInput(this.__wbg_ptr, isLikeNone(value) ? 0xFFFFFF : value ? 1 : 0);
+    }
+    /**
+     * @param {boolean | null} [value]
+     */
+    set supportsAudioOutput(value) {
+        wasm.wasmmodelinfo_set_supportsAudioOutput(this.__wbg_ptr, isLikeNone(value) ? 0xFFFFFF : value ? 1 : 0);
+    }
+    /**
+     * @param {boolean | null} [value]
+     */
+    set supportsFunctionCalling(value) {
+        wasm.wasmmodelinfo_set_supportsFunctionCalling(this.__wbg_ptr, isLikeNone(value) ? 0xFFFFFF : value ? 1 : 0);
+    }
+    /**
+     * @param {boolean | null} [value]
+     */
+    set supportsPromptCaching(value) {
+        wasm.wasmmodelinfo_set_supportsPromptCaching(this.__wbg_ptr, isLikeNone(value) ? 0xFFFFFF : value ? 1 : 0);
+    }
+    /**
+     * @param {boolean | null} [value]
+     */
+    set supportsReasoning(value) {
+        wasm.wasmmodelinfo_set_supportsReasoning(this.__wbg_ptr, isLikeNone(value) ? 0xFFFFFF : value ? 1 : 0);
+    }
+    /**
+     * @param {boolean | null} [value]
+     */
+    set supportsStructuredOutput(value) {
+        wasm.wasmmodelinfo_set_supportsStructuredOutput(this.__wbg_ptr, isLikeNone(value) ? 0xFFFFFF : value ? 1 : 0);
+    }
+    /**
+     * @param {boolean | null} [value]
+     */
+    set supportsVision(value) {
+        wasm.wasmmodelinfo_set_supportsVision(this.__wbg_ptr, isLikeNone(value) ? 0xFFFFFF : value ? 1 : 0);
+    }
+    /**
+     * @param {WasmModelTier[]} value
+     */
+    set tiers(value) {
+        const ptr0 = passArrayJsValueToWasm0(value, wasm.__wbindgen_malloc);
+        const len0 = WASM_VECTOR_LEN;
+        wasm.wasmmodelinfo_set_tiers(this.__wbg_ptr, ptr0, len0);
+    }
+    /**
+     * @returns {boolean | undefined}
+     */
+    get supportsAudioInput() {
+        const ret = wasm.wasmmodelinfo_supportsAudioInput(this.__wbg_ptr);
+        return ret === 0xFFFFFF ? undefined : ret !== 0;
+    }
+    /**
+     * @returns {boolean | undefined}
+     */
+    get supportsAudioOutput() {
+        const ret = wasm.wasmmodelinfo_supportsAudioOutput(this.__wbg_ptr);
+        return ret === 0xFFFFFF ? undefined : ret !== 0;
+    }
+    /**
+     * @returns {boolean | undefined}
+     */
+    get supportsFunctionCalling() {
+        const ret = wasm.wasmmodelinfo_supportsFunctionCalling(this.__wbg_ptr);
+        return ret === 0xFFFFFF ? undefined : ret !== 0;
+    }
+    /**
+     * @returns {boolean | undefined}
+     */
+    get supportsPromptCaching() {
+        const ret = wasm.wasmmodelinfo_supportsPromptCaching(this.__wbg_ptr);
+        return ret === 0xFFFFFF ? undefined : ret !== 0;
+    }
+    /**
+     * @returns {boolean | undefined}
+     */
+    get supportsReasoning() {
+        const ret = wasm.wasmmodelinfo_supportsReasoning(this.__wbg_ptr);
+        return ret === 0xFFFFFF ? undefined : ret !== 0;
+    }
+    /**
+     * @returns {boolean | undefined}
+     */
+    get supportsStructuredOutput() {
+        const ret = wasm.wasmmodelinfo_supportsStructuredOutput(this.__wbg_ptr);
+        return ret === 0xFFFFFF ? undefined : ret !== 0;
+    }
+    /**
+     * @returns {boolean | undefined}
+     */
+    get supportsVision() {
+        const ret = wasm.wasmmodelinfo_supportsVision(this.__wbg_ptr);
+        return ret === 0xFFFFFF ? undefined : ret !== 0;
+    }
+    /**
+     * @returns {WasmModelTier[]}
+     */
+    get tiers() {
+        const ret = wasm.wasmmodelinfo_tiers(this.__wbg_ptr);
+        var v1 = getArrayJsValueFromWasm0(ret[0], ret[1]).slice();
+        wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
+        return v1;
+    }
+}
+if (Symbol.dispose) WasmModelInfo.prototype[Symbol.dispose] = WasmModelInfo.prototype.free;
+
 /**
  * A model available from the API.
  */
@@ -6005,6 +6447,163 @@ export class WasmModelObject {
 }
 if (Symbol.dispose) WasmModelObject.prototype[Symbol.dispose] = WasmModelObject.prototype.free;
 
+/**
+ * Public, FFI-friendly snapshot of a single context-window pricing tier,
+ * projected from `PricingTier`.
+ */
+export class WasmModelTier {
+    static __wrap(ptr) {
+        const obj = Object.create(WasmModelTier.prototype);
+        obj.__wbg_ptr = ptr;
+        WasmModelTierFinalization.register(obj, obj.__wbg_ptr, obj);
+        return obj;
+    }
+    static __unwrap(jsValue) {
+        if (!(jsValue instanceof WasmModelTier)) {
+            return 0;
+        }
+        return jsValue.__destroy_into_raw();
+    }
+    __destroy_into_raw() {
+        const ptr = this.__wbg_ptr;
+        this.__wbg_ptr = 0;
+        WasmModelTierFinalization.unregister(this);
+        return ptr;
+    }
+    free() {
+        const ptr = this.__destroy_into_raw();
+        wasm.__wbg_wasmmodeltier_free(ptr, 0);
+    }
+    /**
+     * @returns {number | undefined}
+     */
+    get cacheCreationInputTokenCost() {
+        const ret = wasm.wasmmodeltier_cacheCreationInputTokenCost(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : ret[1];
+    }
+    /**
+     * @returns {number | undefined}
+     */
+    get cacheReadInputTokenCost() {
+        const ret = wasm.wasmmodeltier_cacheReadInputTokenCost(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : ret[1];
+    }
+    /**
+     * @returns {WasmModelTier}
+     */
+    static default() {
+        const ret = wasm.wasmmodeltier_default();
+        return WasmModelTier.__wrap(ret);
+    }
+    /**
+     * @returns {number | undefined}
+     */
+    get inputCostPerAudioToken() {
+        const ret = wasm.wasmmodeltier_inputCostPerAudioToken(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : ret[1];
+    }
+    /**
+     * @returns {number}
+     */
+    get inputCostPerToken() {
+        const ret = wasm.wasmmodeltier_inputCostPerToken(this.__wbg_ptr);
+        return ret;
+    }
+    /**
+     * @returns {bigint}
+     */
+    get minContextTokens() {
+        const ret = wasm.wasmmodeltier_minContextTokens(this.__wbg_ptr);
+        return BigInt.asUintN(64, ret);
+    }
+    /**
+     * @param {bigint | null} [minContextTokens]
+     * @param {number | null} [inputCostPerToken]
+     * @param {number | null} [outputCostPerToken]
+     * @param {number | null} [cacheReadInputTokenCost]
+     * @param {number | null} [cacheCreationInputTokenCost]
+     * @param {number | null} [inputCostPerAudioToken]
+     * @param {number | null} [outputCostPerAudioToken]
+     * @param {number | null} [outputCostPerReasoningToken]
+     */
+    constructor(minContextTokens, inputCostPerToken, outputCostPerToken, cacheReadInputTokenCost, cacheCreationInputTokenCost, inputCostPerAudioToken, outputCostPerAudioToken, outputCostPerReasoningToken) {
+        const ret = wasm.wasmmodeltier_new(!isLikeNone(minContextTokens), isLikeNone(minContextTokens) ? BigInt(0) : minContextTokens, !isLikeNone(inputCostPerToken), isLikeNone(inputCostPerToken) ? 0 : inputCostPerToken, !isLikeNone(outputCostPerToken), isLikeNone(outputCostPerToken) ? 0 : outputCostPerToken, !isLikeNone(cacheReadInputTokenCost), isLikeNone(cacheReadInputTokenCost) ? 0 : cacheReadInputTokenCost, !isLikeNone(cacheCreationInputTokenCost), isLikeNone(cacheCreationInputTokenCost) ? 0 : cacheCreationInputTokenCost, !isLikeNone(inputCostPerAudioToken), isLikeNone(inputCostPerAudioToken) ? 0 : inputCostPerAudioToken, !isLikeNone(outputCostPerAudioToken), isLikeNone(outputCostPerAudioToken) ? 0 : outputCostPerAudioToken, !isLikeNone(outputCostPerReasoningToken), isLikeNone(outputCostPerReasoningToken) ? 0 : outputCostPerReasoningToken);
+        this.__wbg_ptr = ret;
+        WasmModelTierFinalization.register(this, this.__wbg_ptr, this);
+        return this;
+    }
+    /**
+     * @returns {number | undefined}
+     */
+    get outputCostPerAudioToken() {
+        const ret = wasm.wasmmodeltier_outputCostPerAudioToken(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : ret[1];
+    }
+    /**
+     * @returns {number | undefined}
+     */
+    get outputCostPerReasoningToken() {
+        const ret = wasm.wasmmodeltier_outputCostPerReasoningToken(this.__wbg_ptr);
+        return ret[0] === 0 ? undefined : ret[1];
+    }
+    /**
+     * @returns {number}
+     */
+    get outputCostPerToken() {
+        const ret = wasm.wasmmodeltier_outputCostPerToken(this.__wbg_ptr);
+        return ret;
+    }
+    /**
+     * @param {number | null} [value]
+     */
+    set cacheCreationInputTokenCost(value) {
+        wasm.wasmmodeltier_set_cacheCreationInputTokenCost(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? 0 : value);
+    }
+    /**
+     * @param {number | null} [value]
+     */
+    set cacheReadInputTokenCost(value) {
+        wasm.wasmmodeltier_set_cacheReadInputTokenCost(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? 0 : value);
+    }
+    /**
+     * @param {number | null} [value]
+     */
+    set inputCostPerAudioToken(value) {
+        wasm.wasmmodeltier_set_inputCostPerAudioToken(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? 0 : value);
+    }
+    /**
+     * @param {number} value
+     */
+    set inputCostPerToken(value) {
+        wasm.wasmmodeltier_set_inputCostPerToken(this.__wbg_ptr, value);
+    }
+    /**
+     * @param {bigint} value
+     */
+    set minContextTokens(value) {
+        wasm.wasmmodeltier_set_minContextTokens(this.__wbg_ptr, value);
+    }
+    /**
+     * @param {number | null} [value]
+     */
+    set outputCostPerAudioToken(value) {
+        wasm.wasmmodeltier_set_outputCostPerAudioToken(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? 0 : value);
+    }
+    /**
+     * @param {number | null} [value]
+     */
+    set outputCostPerReasoningToken(value) {
+        wasm.wasmmodeltier_set_outputCostPerReasoningToken(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? 0 : value);
+    }
+    /**
+     * @param {number} value
+     */
+    set outputCostPerToken(value) {
+        wasm.wasmmodeltier_set_outputCostPerToken(this.__wbg_ptr, value);
+    }
+}
+if (Symbol.dispose) WasmModelTier.prototype[Symbol.dispose] = WasmModelTier.prototype.free;
+
 /**
  * Response listing available models.
  */
@@ -7820,6 +8419,16 @@ export const WasmReasoningEffort = Object.freeze({
     High: 2, "2": "High",
 });
 
+/**
+ * Result of a `refresh_catalog` call.
+ * @enum {0 | 1 | 2}
+ */
+export const WasmRefreshOutcome = Object.freeze({
+    Disabled: 0, "0": "Disabled",
+    FromCache: 1, "1": "FromCache",
+    Fetched: 2, "2": "Fetched",
+});
+
 /**
  * A document to be reranked — either a plain string or an object with a text field.
  * @enum {0 | 1}
@@ -10714,7 +11323,7 @@ export function allProviders() {
 /**
  * Return the capability flags for a named provider.
  *
- * Performs an O(n) linear scan over the embedded registry (143 entries).
+ * Performs an O(n) linear scan over the embedded registry (163 entries).
  * Returns an owned value so bindings can pass capability data without
  * borrowing registry internals.
  *
@@ -10765,6 +11374,18 @@ export function clear() {
     wasm.clear();
 }
 
+/**
+ * Clear the overlay registry, reverting `completion_cost`,
+ * `completion_cost_with_cache`, and `model_info` to the
+ * embedded catalog.
+ *
+ * Primarily a test seam (see `install_catalog_overlay_from_str`); also
+ * usable by long-running processes that want to abandon a runtime refresh.
+ */
+export function clearCatalogOverlay() {
+    wasm.clearCatalogOverlay();
+}
+
 /**
  * Calculate the estimated cost of a completion given a model name and token
  * counts.
@@ -10802,6 +11423,12 @@ export function completionCost(model, prompt_tokens, completion_tokens) {
  *
  * Returns `None` if the model is not present in the embedded pricing
  * registry, mirroring `completion_cost`.
+ *
+ * When the model has `ModelPricing.tiers`, the tier whose
+ * `min_context_tokens` is the highest value `<= prompt_tokens` supplies the
+ * input/output/cache rates for the whole call; models without tiers (or
+ * when `prompt_tokens` is below every tier threshold) use the base rates
+ * unchanged, matching the original flat-rate behaviour.
  * @param {string} model
  * @param {bigint} prompt_tokens
  * @param {bigint} cached_tokens
@@ -10937,6 +11564,82 @@ export function encodeDataUrl(bytes, mime) {
     }
 }
 
+/**
+ * Install the overlay registry from a raw catalog JSON string, bypassing
+ * the network and disk cache entirely.
+ *
+ * Parses and flattens `catalog_json` with the same
+ * `registry_from_catalog_str` logic used for the embedded catalog and the
+ * network refresh path, then atomically swaps it in as the active overlay.
+ * A parse failure returns `CatalogRefreshError.Parse` and leaves any
+ * existing overlay untouched.
+ *
+ * This is primarily a testable seam: it lets tests exercise overlay
+ * installation and the embedded/overlay fallback behavior in
+ * `completion_cost` / `model_info` without a real network
+ * call.
+ * @param {string} catalog_json
+ */
+export function installCatalogOverlayFromStr(catalog_json) {
+    const ptr0 = passStringToWasm0(catalog_json, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.installCatalogOverlayFromStr(ptr0, len0);
+    if (ret[1]) {
+        throw takeFromExternrefTable0(ret[0]);
+    }
+}
+
+/**
+ * Look up FFI-friendly pricing and capability metadata for a model.
+ *
+ * Returns `None` if the model is not present in the active pricing
+ * registry. Uses the same exact-match-then-prefix-fallback resolution as
+ * `model_pricing`; unlike `model_pricing`, the result is an owned
+ * `ModelInfo` value safe to hand across the FFI boundary.
+ *
+ * When a runtime catalog refresh has succeeded, this reflects the refreshed
+ * (overlay) catalog; otherwise it reflects the embedded catalog. See
+ * `model_pricing` for the embedded-only alternative.
+ * @param {string} model
+ * @returns {WasmModelInfo | undefined}
+ */
+export function modelInfo(model) {
+    const ptr0 = passStringToWasm0(model, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.modelInfo(ptr0, len0);
+    return ret === 0 ? undefined : WasmModelInfo.__wrap(ret);
+}
+
+/**
+ * Refresh the runtime catalog overlay per `config`.
+ *
+ * - `config.enabled == false`: returns `Ok(``RefreshOutcome.Disabled``)`
+ *   immediately. No network, filesystem, or overlay activity.
+ * - A fresh on-disk cache (age < `config.ttl_seconds`) exists at the
+ *   resolved cache path (`config.cache_path`, or a default under
+ *   `std.env.temp_dir()`): read + flatten it and install the overlay,
+ *   returning `Ok(``RefreshOutcome.FromCache``)`. No network request is
+ *   made.
+ * - Otherwise: validate `config.source_url` uses `https`
+ *   (`CatalogRefreshError.InsecureUrl` otherwise), fetch it, flatten it,
+ *   install the overlay, best-effort write the raw JSON to the cache path
+ *   (a cache write failure does not fail the refresh), and return
+ *   `Ok(``RefreshOutcome.Fetched``)`.
+ *
+ * On any error return, the overlay is left untouched: the previously
+ * active registry (a prior successful overlay, or the embedded catalog if
+ * none was ever installed) remains in effect. This is what makes the
+ * feature air-gap-safe — an unreachable or invalid `source_url` never
+ * degrades `completion_cost` / `model_info` below embedded-catalog
+ * availability.
+ * @param {any} config
+ * @returns {Promise<WasmRefreshOutcome>}
+ */
+export function refreshCatalog(config) {
+    const ret = wasm.refreshCatalog(config);
+    return ret;
+}
+
 /**
  * Register a custom provider in the global runtime registry.
  *
@@ -10986,6 +11689,17 @@ function __wbg_get_imports() {
             const ret = Error(getStringFromWasm0(arg0, arg1));
             return ret;
         },
+        __wbg_Number_6b506e6536831eaa: function(arg0) {
+            const ret = Number(arg0);
+            return ret;
+        },
+        __wbg_String_8564e559799eccda: function(arg0, arg1) {
+            const ret = String(arg1);
+            const ptr1 = passStringToWasm0(ret, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+            const len1 = WASM_VECTOR_LEN;
+            getDataViewMemory0().setInt32(arg0 + 4 * 1, len1, true);
+            getDataViewMemory0().setInt32(arg0 + 4 * 0, ptr1, true);
+        },
         __wbg___wbindgen_bigint_get_as_i64_38130e98eecd467d: function(arg0, arg1) {
             const v = arg1;
             const ret = typeof(v) === 'bigint' ? v : undefined;
@@ -11180,6 +11894,10 @@ function __wbg_get_imports() {
             const ret = arg0.value;
             return ret;
         },
+        __wbg_get_with_ref_key_6412cf3094599694: function(arg0, arg1) {
+            const ret = arg0[arg1];
+            return ret;
+        },
         __wbg_has_73740b27f436fed3: function() { return handleError(function (arg0, arg1) {
             const ret = Reflect.has(arg0, arg1);
             return ret;
@@ -11554,6 +12272,14 @@ function __wbg_get_imports() {
             const ret = WasmModelsListResponse.__wrap(arg0);
             return ret;
         },
+        __wbg_wasmmodeltier_new: function(arg0) {
+            const ret = WasmModelTier.__wrap(arg0);
+            return ret;
+        },
+        __wbg_wasmmodeltier_unwrap: function(arg0) {
+            const ret = WasmModelTier.__unwrap(arg0);
+            return ret;
+        },
         __wbg_wasmmoderationresponse_new: function(arg0) {
             const ret = WasmModerationResponse.__wrap(arg0);
             return ret;
@@ -11671,22 +12397,22 @@ function __wbg_get_imports() {
             return ret;
         },
         __wbindgen_cast_0000000000000001: function(arg0, arg1) {
-            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 759, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
+            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 774, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
             const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h838b8ae4822e18a5);
             return ret;
         },
         __wbindgen_cast_0000000000000002: function(arg0, arg1) {
-            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 779, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
+            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 794, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
             const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__hb42333e27c827052);
             return ret;
         },
         __wbindgen_cast_0000000000000003: function(arg0, arg1) {
-            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 684, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
+            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 699, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
             const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h238043eefd9b3f4d);
             return ret;
         },
         __wbindgen_cast_0000000000000004: function(arg0, arg1) {
-            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 700, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
+            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 715, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
             const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__ha49c00e409183b38);
             return ret;
         },
@@ -11803,6 +12529,9 @@ const WasmBatchObjectFinalization = (typeof FinalizationRegistry === 'undefined'
 const WasmBatchRequestCountsFinalization = (typeof FinalizationRegistry === 'undefined')
     ? { register: () => {}, unregister: () => {} }
     : new FinalizationRegistry(ptr => wasm.__wbg_wasmbatchrequestcounts_free(ptr, 1));
+const WasmCatalogRefreshConfigFinalization = (typeof FinalizationRegistry === 'undefined')
+    ? { register: () => {}, unregister: () => {} }
+    : new FinalizationRegistry(ptr => wasm.__wbg_wasmcatalogrefreshconfig_free(ptr, 1));
 const WasmChatCompletionChunkFinalization = (typeof FinalizationRegistry === 'undefined')
     ? { register: () => {}, unregister: () => {} }
     : new FinalizationRegistry(ptr => wasm.__wbg_wasmchatcompletionchunk_free(ptr, 1));
@@ -11902,9 +12631,15 @@ const WasmLiterLlmErrorFinalization = (typeof FinalizationRegistry === 'undefine
 const WasmMessageFinalization = (typeof FinalizationRegistry === 'undefined')
     ? { register: () => {}, unregister: () => {} }
     : new FinalizationRegistry(ptr => wasm.__wbg_wasmmessage_free(ptr, 1));
+const WasmModelInfoFinalization = (typeof FinalizationRegistry === 'undefined')
+    ? { register: () => {}, unregister: () => {} }
+    : new FinalizationRegistry(ptr => wasm.__wbg_wasmmodelinfo_free(ptr, 1));
 const WasmModelObjectFinalization = (typeof FinalizationRegistry === 'undefined')
     ? { register: () => {}, unregister: () => {} }
     : new FinalizationRegistry(ptr => wasm.__wbg_wasmmodelobject_free(ptr, 1));
+const WasmModelTierFinalization = (typeof FinalizationRegistry === 'undefined')
+    ? { register: () => {}, unregister: () => {} }
+    : new FinalizationRegistry(ptr => wasm.__wbg_wasmmodeltier_free(ptr, 1));
 const WasmModelsListResponseFinalization = (typeof FinalizationRegistry === 'undefined')
     ? { register: () => {}, unregister: () => {} }
     : new FinalizationRegistry(ptr => wasm.__wbg_wasmmodelslistresponse_free(ptr, 1));

package/package.json

@@ -2,7 +2,7 @@
   "name": "@xberg-io/liter-llm-wasm",
   "type": "module",
   "description": "Universal LLM API client with Rust-powered polyglot bindings.",
-  "version": "1.9.3",
+  "version": "1.10.0",
   "license": "MIT",
   "repository": {
     "type": "git",