ai-world-sdk 1.5.11 → 1.5.13

package/pacakges/ai-sdk-provider/src/types/openrouter-chat-settings.ts

@@ -1,233 +0,0 @@
-import type { OpenRouterSharedSettings } from '..';
-import type {
-  DataCollection,
-  Engine,
-  IdAutoRouter,
-  IdFileParser,
-  IdModeration,
-  IdResponseHealing,
-  IdWeb,
-  PdfEngine,
-  ProviderSort,
-  Quantization,
-} from './openrouter-api-types';
-
-// https://openrouter.ai/api/v1/models
-export type OpenRouterChatModelId = string;
-
-export type OpenRouterChatSettings = {
-  /**
-Modify the likelihood of specified tokens appearing in the completion.
-
-Accepts a JSON object that maps tokens (specified by their token ID in
-the GPT tokenizer) to an associated bias value from -100 to 100. You
-can use this tokenizer tool to convert text to token IDs. Mathematically,
-the bias is added to the logits generated by the model prior to sampling.
-The exact effect will vary per model, but values between -1 and 1 should
-decrease or increase likelihood of selection; values like -100 or 100
-should result in a ban or exclusive selection of the relevant token.
-
-As an example, you can pass {"50256": -100} to prevent the <|endoftext|>
-token from being generated.
-*/
-  logitBias?: Record<number, number>;
-
-  /**
-Return the log probabilities of the tokens. Including logprobs will increase
-the response size and can slow down response times. However, it can
-be useful to understand better how the model is behaving.
-
-Setting to true will return the log probabilities of the tokens that
-were generated.
-
-Setting to a number will return the log probabilities of the top n
-tokens that were generated.
-*/
-  logprobs?: boolean | number;
-
-  /**
-Whether to enable parallel function calling during tool use. Default to true.
-   */
-  parallelToolCalls?: boolean;
-
-  /**
-A unique identifier representing your end-user, which can help OpenRouter to
-monitor and detect abuse. Learn more.
-*/
-  user?: string;
-
-  /**
-   * Plugin configurations for enabling various capabilities
-   */
-  plugins?: Array<
-    | {
-        id: IdWeb;
-        max_results?: number;
-        search_prompt?: string;
-        engine?: Engine;
-      }
-    | {
-        id: IdFileParser;
-        max_files?: number;
-        pdf?: {
-          engine?: PdfEngine;
-        };
-      }
-    | {
-        id: IdModeration;
-      }
-    | {
-        /**
-         * Response healing plugin - automatically validates and repairs malformed JSON responses.
-         *
-         * **Important:** This plugin only works with non-streaming requests (e.g., `generateObject`).
-         * It has no effect when used with streaming methods like `streamObject` or `streamText`.
-         * The plugin activates when using `response_format` with `json_schema` or `json_object`.
-         *
-         * @see https://openrouter.ai/docs/guides/features/plugins/response-healing
-         */
-        id: IdResponseHealing;
-      }
-    | {
-        /**
-         * Auto-router plugin - configures allowed models when using `openrouter/auto`.
-         *
-         * Use wildcard patterns to restrict which models the auto router can select from.
-         * When no `allowed_models` are specified, the auto router uses all supported models.
-         *
-         * @see https://openrouter.ai/docs/guides/routing/routers/auto-router
-         */
-        id: IdAutoRouter;
-        allowed_models?: string[];
-      }
-  >;
-
-  /**
-   * Built-in web search options for models that support native web search
-   */
-  web_search_options?: {
-    /**
-     * Maximum number of search results to include
-     */
-    max_results?: number;
-    /**
-     * Custom search prompt to guide the search query
-     */
-    search_prompt?: string;
-    /**
-     * Search engine to use for web search
-     * - "native": Use provider's built-in web search
-     * - "exa": Use Exa's search API
-     * - undefined: Native if supported, otherwise Exa
-     * @see https://openrouter.ai/docs/features/web-search
-     */
-    engine?: Engine;
-  };
-
-  /**
-   * Enable Anthropic automatic prompt caching by setting a top-level cache_control
-   * directive on the request body. When set to `{ type: 'ephemeral' }`, Anthropic
-   * will automatically cache eligible content in your prompts.
-   *
-   * Only works with Anthropic models through OpenRouter.
-   *
-   * @see https://platform.claude.com/docs/en/build-with-claude/prompt-caching#automatic-caching
-   * @see https://openrouter.ai/docs
-   */
-  cache_control?: {
-    type: 'ephemeral';
-    /**
-     * Optional time-to-live for the cache entry.
-     * - `'5m'` — 5 minutes (default when omitted)
-     * - `'1h'` — 1 hour
-     */
-    ttl?: '5m' | '1h';
-  };
-
-  /**
-   * Debug options for troubleshooting API requests.
-   * Only works with streaming requests.
-   * @see https://openrouter.ai/docs/api-reference/debugging
-   */
-  debug?: {
-    /**
-     * When true, echoes back the request body that was sent to the upstream provider.
-     * The debug data will be returned as the first chunk in the stream with a `debug.echo_upstream_body` field.
-     * Sensitive data like user IDs and base64 content will be redacted.
-     */
-    echo_upstream_body?: boolean;
-  };
-
-  /**
-   * Structured-output options forwarded to OpenRouter's
-   * `response_format.json_schema` payload.
-   *
-   * Use this to opt out of strict mode for models whose providers don't
-   * advertise support for it (e.g. open-source models routed through
-   * non-OpenAI-compatible providers). When `strict` is left unset, the
-   * SDK defaults to `true` for backward compatibility.
-   *
-   * Only applies when a `responseFormat` with a `schema` is provided on
-   * the call site. Has no effect for `json_object` or text responses.
-   */
-  structuredOutputs?: {
-    /**
-     * Whether to set `response_format.json_schema.strict` on the outbound
-     * request. Defaults to `true` when omitted.
-     */
-    strict?: boolean;
-  };
-
-  /**
-   * Provider routing preferences to control request routing behavior
-   */
-  provider?: {
-    /**
-     * List of provider slugs to try in order (e.g. ["anthropic", "openai"])
-     */
-    order?: string[];
-    /**
-     * Whether to allow backup providers when primary is unavailable (default: true)
-     */
-    allow_fallbacks?: boolean;
-    /**
-     * Only use providers that support all parameters in your request (default: false)
-     */
-    require_parameters?: boolean;
-    /**
-     * Control whether to use providers that may store data
-     */
-    data_collection?: DataCollection;
-    /**
-     * List of provider slugs to allow for this request
-     */
-    only?: string[];
-    /**
-     * List of provider slugs to skip for this request
-     */
-    ignore?: string[];
-    /**
-     * List of quantization levels to filter by (e.g. ["int4", "int8"])
-     */
-    quantizations?: Array<Quantization>;
-    /**
-     * Sort providers by price, throughput, or latency
-     */
-    sort?: ProviderSort;
-    /**
-     * Maximum pricing you want to pay for this request
-     */
-    max_price?: {
-      prompt?: number | string;
-      completion?: number | string;
-      image?: number | string;
-      audio?: number | string;
-      request?: number | string;
-    };
-    /**
-     * Whether to restrict routing to only ZDR (Zero Data Retention) endpoints.
-     * When true, only endpoints that do not retain prompts will be used.
-     */
-    zdr?: boolean;
-  };
-} & OpenRouterSharedSettings;

package/pacakges/ai-sdk-provider/src/types/openrouter-completion-settings.ts

@@ -1,39 +0,0 @@
-import type { OpenRouterSharedSettings } from '.';
-
-export type OpenRouterCompletionModelId = string;
-
-export type OpenRouterCompletionSettings = {
-  /**
-Modify the likelihood of specified tokens appearing in the completion.
-
-Accepts a JSON object that maps tokens (specified by their token ID in
-the GPT tokenizer) to an associated bias value from -100 to 100. You
-can use this tokenizer tool to convert text to token IDs. Mathematically,
-the bias is added to the logits generated by the model prior to sampling.
-The exact effect will vary per model, but values between -1 and 1 should
-decrease or increase likelihood of selection; values like -100 or 100
-should result in a ban or exclusive selection of the relevant token.
-
-As an example, you can pass {"50256": -100} to prevent the <|endoftext|>
-token from being generated.
-   */
-  logitBias?: Record<number, number>;
-
-  /**
-Return the log probabilities of the tokens. Including logprobs will increase
-the response size and can slow down response times. However, it can
-be useful to better understand how the model is behaving.
-
-Setting to true will return the log probabilities of the tokens that
-were generated.
-
-Setting to a number will return the log probabilities of the top n
-tokens that were generated.
-   */
-  logprobs?: boolean | number;
-
-  /**
-The suffix that comes after a completion of inserted text.
-   */
-  suffix?: string;
-} & OpenRouterSharedSettings;

package/pacakges/ai-sdk-provider/src/types/openrouter-embedding-settings.ts

@@ -1,56 +0,0 @@
-import type { OpenRouterSharedSettings } from '..';
-
-// https://openrouter.ai/api/v1/models
-export type OpenRouterEmbeddingModelId = string;
-
-export type OpenRouterEmbeddingSettings = {
-  /**
-   * A unique identifier representing your end-user, which can help OpenRouter to
-   * monitor and detect abuse.
-   */
-  user?: string;
-
-  /**
-   * Provider routing preferences to control request routing behavior
-   */
-  provider?: {
-    /**
-     * List of provider slugs to try in order (e.g. ["openai", "voyageai"])
-     */
-    order?: string[];
-    /**
-     * Whether to allow backup providers when primary is unavailable (default: true)
-     */
-    allow_fallbacks?: boolean;
-    /**
-     * Only use providers that support all parameters in your request (default: false)
-     */
-    require_parameters?: boolean;
-    /**
-     * Control whether to use providers that may store data
-     */
-    data_collection?: 'allow' | 'deny';
-    /**
-     * List of provider slugs to allow for this request
-     */
-    only?: string[];
-    /**
-     * List of provider slugs to skip for this request
-     */
-    ignore?: string[];
-    /**
-     * Sort providers by price, throughput, or latency
-     */
-    sort?: 'price' | 'throughput' | 'latency';
-    /**
-     * Maximum pricing you want to pay for this request
-     */
-    max_price?: {
-      prompt?: number | string;
-      completion?: number | string;
-      image?: number | string;
-      audio?: number | string;
-      request?: number | string;
-    };
-  };
-} & OpenRouterSharedSettings;

package/pacakges/ai-sdk-provider/src/types/openrouter-image-settings.ts

@@ -1,49 +0,0 @@
-import type { OpenRouterSharedSettings } from '..';
-
-// https://openrouter.ai/api/v1/models
-export type OpenRouterImageModelId = string;
-
-export type OpenRouterImageSettings = {
-  /**
-   * Provider routing preferences to control request routing behavior
-   */
-  provider?: {
-    /**
-     * List of provider slugs to try in order (e.g. ["google", "black-forest-labs"])
-     */
-    order?: string[];
-    /**
-     * Whether to allow backup providers when primary is unavailable (default: true)
-     */
-    allow_fallbacks?: boolean;
-    /**
-     * Only use providers that support all parameters in your request (default: false)
-     */
-    require_parameters?: boolean;
-    /**
-     * Control whether to use providers that may store data
-     */
-    data_collection?: 'allow' | 'deny';
-    /**
-     * List of provider slugs to allow for this request
-     */
-    only?: string[];
-    /**
-     * List of provider slugs to skip for this request
-     */
-    ignore?: string[];
-    /**
-     * Sort providers by price, throughput, or latency
-     */
-    sort?: 'price' | 'throughput' | 'latency';
-    /**
-     * Maximum pricing you want to pay for this request
-     */
-    max_price?: {
-      prompt?: number | string;
-      completion?: number | string;
-      image?: number | string;
-      request?: number | string;
-    };
-  };
-} & OpenRouterSharedSettings;

package/pacakges/ai-sdk-provider/src/types/openrouter-video-settings.ts

@@ -1,26 +0,0 @@
-export type OpenRouterVideoModelId = string;
-
-export type OpenRouterVideoSettings = {
-  /**
-   * Whether to generate audio alongside the video.
-   * Defaults to the endpoint's generate_audio capability flag, false if not set.
-   */
-  generateAudio?: boolean;
-
-  /**
-   * Polling interval in milliseconds when waiting for video generation to complete.
-   * @default 2000
-   */
-  pollIntervalMs?: number;
-
-  /**
-   * Maximum time in milliseconds to wait for video generation to complete.
-   * @default 600000 (10 minutes)
-   */
-  maxPollTimeMs?: number;
-
-  /**
-   * Additional body parameters to send with the video generation request.
-   */
-  extraBody?: Record<string, unknown>;
-};

package/pacakges/ai-sdk-provider/src/utils/compute-token-usage.test.ts

@@ -1,186 +0,0 @@
-import { describe, expect, it } from 'vitest';
-import { computeTokenUsage, emptyUsage } from './compute-token-usage';
-
-describe('computeTokenUsage', () => {
-  it('should compute all fields from complete usage data', () => {
-    const result = computeTokenUsage({
-      prompt_tokens: 100,
-      completion_tokens: 50,
-      prompt_tokens_details: {
-        cached_tokens: 20,
-        cache_write_tokens: 30,
-      },
-      completion_tokens_details: {
-        reasoning_tokens: 10,
-      },
-    });
-
-    expect(result.inputTokens).toStrictEqual({
-      total: 100,
-      noCache: 80,
-      cacheRead: 20,
-      cacheWrite: 30,
-    });
-    expect(result.outputTokens).toStrictEqual({
-      total: 50,
-      text: 40,
-      reasoning: 10,
-    });
-    expect(result.raw).toBeDefined();
-  });
-
-  it('should set noCache equal to total when no detail fields present', () => {
-    const result = computeTokenUsage({
-      prompt_tokens: 50,
-      completion_tokens: 30,
-    });
-
-    expect(result.inputTokens).toStrictEqual({
-      total: 50,
-      noCache: 50,
-      cacheRead: 0,
-      cacheWrite: undefined,
-    });
-    expect(result.outputTokens).toStrictEqual({
-      total: 30,
-      text: 30,
-      reasoning: 0,
-    });
-  });
-
-  it('should handle null prompt_tokens_details', () => {
-    const result = computeTokenUsage({
-      prompt_tokens: 10,
-      completion_tokens: 5,
-      prompt_tokens_details: null,
-      completion_tokens_details: null,
-    });
-
-    expect(result.inputTokens).toStrictEqual({
-      total: 10,
-      noCache: 10,
-      cacheRead: 0,
-      cacheWrite: undefined,
-    });
-    expect(result.outputTokens).toStrictEqual({
-      total: 5,
-      text: 5,
-      reasoning: 0,
-    });
-  });
-
-  it('should handle null cache_write_tokens vs undefined vs number', () => {
-    const withNull = computeTokenUsage({
-      prompt_tokens: 10,
-      completion_tokens: 5,
-      prompt_tokens_details: { cached_tokens: 0, cache_write_tokens: null },
-    });
-    expect(withNull.inputTokens.cacheWrite).toBeUndefined();
-
-    const withUndefined = computeTokenUsage({
-      prompt_tokens: 10,
-      completion_tokens: 5,
-      prompt_tokens_details: {
-        cached_tokens: 0,
-        cache_write_tokens: undefined,
-      },
-    });
-    expect(withUndefined.inputTokens.cacheWrite).toBeUndefined();
-
-    const withNumber = computeTokenUsage({
-      prompt_tokens: 10,
-      completion_tokens: 5,
-      prompt_tokens_details: { cached_tokens: 0, cache_write_tokens: 42 },
-    });
-    expect(withNumber.inputTokens.cacheWrite).toBe(42);
-  });
-
-  it('should handle zero token counts', () => {
-    const result = computeTokenUsage({
-      prompt_tokens: 0,
-      completion_tokens: 0,
-      prompt_tokens_details: { cached_tokens: 0 },
-      completion_tokens_details: { reasoning_tokens: 0 },
-    });
-
-    expect(result.inputTokens).toStrictEqual({
-      total: 0,
-      noCache: 0,
-      cacheRead: 0,
-      cacheWrite: undefined,
-    });
-    expect(result.outputTokens).toStrictEqual({
-      total: 0,
-      text: 0,
-      reasoning: 0,
-    });
-  });
-
-  it('should handle missing prompt_tokens and completion_tokens', () => {
-    const result = computeTokenUsage({});
-
-    expect(result.inputTokens).toStrictEqual({
-      total: 0,
-      noCache: 0,
-      cacheRead: 0,
-      cacheWrite: undefined,
-    });
-    expect(result.outputTokens).toStrictEqual({
-      total: 0,
-      text: 0,
-      reasoning: 0,
-    });
-  });
-
-  it('should handle null prompt_tokens and completion_tokens', () => {
-    const result = computeTokenUsage({
-      prompt_tokens: null,
-      completion_tokens: null,
-    });
-
-    expect(result.inputTokens.total).toBe(0);
-    expect(result.outputTokens.total).toBe(0);
-  });
-
-  it('should preserve raw usage object', () => {
-    const usage = {
-      prompt_tokens: 10,
-      completion_tokens: 5,
-      total_tokens: 15,
-      extra_field: 'preserved',
-    };
-
-    const result = computeTokenUsage(usage);
-    expect(result.raw).toBeDefined();
-    expect((result.raw as Record<string, unknown>).total_tokens).toBe(15);
-    expect((result.raw as Record<string, unknown>).extra_field).toBe(
-      'preserved',
-    );
-  });
-});
-
-describe('emptyUsage', () => {
-  it('should return undefined for all detail fields', () => {
-    const result = emptyUsage();
-
-    expect(result.inputTokens).toStrictEqual({
-      total: 0,
-      noCache: undefined,
-      cacheRead: undefined,
-      cacheWrite: undefined,
-    });
-    expect(result.outputTokens).toStrictEqual({
-      total: 0,
-      text: undefined,
-      reasoning: undefined,
-    });
-    expect(result.raw).toBeUndefined();
-  });
-
-  it('should return a new object each call', () => {
-    const a = emptyUsage();
-    const b = emptyUsage();
-    expect(a).not.toBe(b);
-    expect(a.inputTokens).not.toBe(b.inputTokens);
-  });
-});

package/pacakges/ai-sdk-provider/src/utils/compute-token-usage.ts

@@ -1,55 +0,0 @@
-import type { JSONObject, LanguageModelV3Usage } from '@ai-sdk/provider';
-
-interface UsageData {
-  prompt_tokens?: number | null;
-  completion_tokens?: number | null;
-  prompt_tokens_details?: {
-    cached_tokens?: number;
-    cache_write_tokens?: number | null;
-  } | null;
-  completion_tokens_details?: {
-    reasoning_tokens?: number;
-  } | null;
-}
-
-export function computeTokenUsage(usage: UsageData): LanguageModelV3Usage {
-  const promptTokens = usage.prompt_tokens ?? 0;
-  const completionTokens = usage.completion_tokens ?? 0;
-  const cacheReadTokens = usage.prompt_tokens_details?.cached_tokens ?? 0;
-  const cacheWriteTokens =
-    usage.prompt_tokens_details?.cache_write_tokens ?? undefined;
-  const reasoningTokens =
-    usage.completion_tokens_details?.reasoning_tokens ?? 0;
-
-  return {
-    inputTokens: {
-      total: promptTokens,
-      noCache: promptTokens - cacheReadTokens,
-      cacheRead: cacheReadTokens,
-      cacheWrite: cacheWriteTokens,
-    },
-    outputTokens: {
-      total: completionTokens,
-      text: completionTokens - reasoningTokens,
-      reasoning: reasoningTokens,
-    },
-    raw: usage as JSONObject,
-  };
-}
-
-export function emptyUsage(): LanguageModelV3Usage {
-  return {
-    inputTokens: {
-      total: 0,
-      noCache: undefined,
-      cacheRead: undefined,
-      cacheWrite: undefined,
-    },
-    outputTokens: {
-      total: 0,
-      text: undefined,
-      reasoning: undefined,
-    },
-    raw: undefined,
-  };
-}