@anthropic-ai/sdk 0.101.0 → 0.103.0

package/src/lib/tools/BetaToolRunner.ts

@@ -199,8 +199,17 @@ export class BetaToolRunner<Stream extends boolean> {
           const isCompacted = await this.#checkAndCompact();
           if (!isCompacted) {
             if (!this.#mutated) {
-              const { role, content } = await this.#message;
-              this.#state.params.messages.push({ role, content });
+              const message = await this.#message;
+              this.#state.params.messages.push({ role: message.role, content: message.content });
+
+              // Refusal-terminated turns are terminal: the refusal may have cut a tool_use off
+              // with partial input, so executing this turn's tools would fire side effects the
+              // model never confirmed — and once middleware strips the refusal turn, their
+              // tool_results could never be replayed coherently. Surface the refusal as the
+              // final message instead.
+              if (message.stop_reason === 'refusal') {
+                break;
+              }
             }
 
             const toolMessage = await this.#generateToolResponse(this.#state.params.messages.at(-1)!);
@@ -521,4 +530,4 @@ export type BetaToolRunnerParams = Simplify<
   }
 >;
 
-export type BetaToolRunnerRequestOptions = Pick<RequestOptions, 'headers' | 'signal'>;
+export type BetaToolRunnerRequestOptions = Pick<RequestOptions, 'headers' | 'signal' | 'fallbackState'>;

package/src/resources/beta/agents/agents.ts

@@ -519,20 +519,13 @@ export interface BetaManagedAgentsCustomTool {
  * JSON Schema for custom tool input parameters.
  */
 export interface BetaManagedAgentsCustomToolInputSchema {
-  /**
-   * JSON Schema properties defining the tool's input parameters.
-   */
+  type: 'object';
+
   properties?: { [key: string]: unknown } | null;
 
-  /**
-   * List of required property names.
-   */
-  required?: Array<string>;
+  required?: Array<string> | null;
 
-  /**
-   * Must be 'object' for tool input schemas.
-   */
-  type?: 'object';
+  [k: string]: unknown;
 }
 
 /**
@@ -677,6 +670,7 @@ export interface BetaManagedAgentsMCPToolsetParams {
  * details and options.
  */
 export type BetaManagedAgentsModel =
+  | 'claude-fable-5'
   | 'claude-opus-4-8'
   | 'claude-opus-4-7'
   | 'claude-opus-4-6'
@@ -835,12 +829,12 @@ export interface AgentCreateParams {
   model: BetaManagedAgentsModel | BetaManagedAgentsModelConfigParams;
 
   /**
-   * Body param: Human-readable name for the agent. 1-256 characters.
+   * Body param: Human-readable name for the agent.
    */
   name: string;
 
   /**
-   * Body param: Description of what the agent does. Up to 2048 characters.
+   * Body param: Description of what the agent does.
    */
   description?: string | null;
 
@@ -864,12 +858,12 @@ export interface AgentCreateParams {
   multiagent?: SessionsAPI.BetaManagedAgentsMultiagentParams | null;
 
   /**
-   * Body param: Skills available to the agent. Maximum 20.
+   * Body param: Skills available to the agent.
    */
   skills?: Array<BetaManagedAgentsSkillParams>;
 
   /**
-   * Body param: System prompt for the agent. Up to 100,000 characters.
+   * Body param: System prompt for the agent.
    */
   system?: string | null;
 
@@ -911,8 +905,7 @@ export interface AgentUpdateParams {
   version: number;
 
   /**
-   * Body param: Description. Up to 2048 characters. Omit to preserve; send empty
-   * string or null to clear.
+   * Body param: Description. Omit to preserve; send empty string or null to clear.
    */
   description?: string | null;
 
@@ -945,20 +938,19 @@ export interface AgentUpdateParams {
   multiagent?: SessionsAPI.BetaManagedAgentsMultiagentParams | null;
 
   /**
-   * Body param: Human-readable name. 1-256 characters. Omit to preserve. Cannot be
+   * Body param: Human-readable name. Must be non-empty. Omit to preserve. Cannot be
    * cleared.
    */
   name?: string;
 
   /**
    * Body param: Skills. Full replacement. Omit to preserve; send empty array or null
-   * to clear. Maximum 20.
+   * to clear.
    */
   skills?: Array<BetaManagedAgentsSkillParams> | null;
 
   /**
-   * Body param: System prompt. Up to 100,000 characters. Omit to preserve; send
-   * empty string or null to clear.
+   * Body param: System prompt. Omit to preserve; send empty string or null to clear.
    */
   system?: string | null;
 

package/src/resources/beta/beta.ts

@@ -239,6 +239,12 @@ import {
   BetaDocumentBlock,
   BetaEncryptedCodeExecutionResultBlock,
   BetaEncryptedCodeExecutionResultBlockParam,
+  BetaFallbackBlock,
+  BetaFallbackBlockParam,
+  BetaFallbackInfo,
+  BetaFallbackInfoParam,
+  BetaFallbackMessageIterationUsage,
+  BetaFallbackParam,
   BetaFileDocumentSource,
   BetaFileImageSource,
   BetaImageBlockParam,
@@ -472,7 +478,9 @@ export type AnthropicBeta =
   | 'advisor-tool-2026-03-01'
   | 'managed-agents-2026-04-01'
   | 'cache-diagnosis-2026-04-07'
-  | 'thinking-token-count-2026-05-13';
+  | 'thinking-token-count-2026-05-13'
+  | 'server-side-fallback-2026-06-01'
+  | 'fallback-credit-2026-06-01';
 
 export interface BetaAPIError {
   message: string;
@@ -673,6 +681,12 @@ export declare namespace Beta {
     type BetaDocumentBlock as BetaDocumentBlock,
     type BetaEncryptedCodeExecutionResultBlock as BetaEncryptedCodeExecutionResultBlock,
     type BetaEncryptedCodeExecutionResultBlockParam as BetaEncryptedCodeExecutionResultBlockParam,
+    type BetaFallbackBlock as BetaFallbackBlock,
+    type BetaFallbackBlockParam as BetaFallbackBlockParam,
+    type BetaFallbackInfo as BetaFallbackInfo,
+    type BetaFallbackInfoParam as BetaFallbackInfoParam,
+    type BetaFallbackMessageIterationUsage as BetaFallbackMessageIterationUsage,
+    type BetaFallbackParam as BetaFallbackParam,
     type BetaFileDocumentSource as BetaFileDocumentSource,
     type BetaFileImageSource as BetaFileImageSource,
     type BetaImageBlockParam as BetaImageBlockParam,

package/src/resources/beta/index.ts

@@ -192,6 +192,12 @@ export {
   type BetaDocumentBlock,
   type BetaEncryptedCodeExecutionResultBlock,
   type BetaEncryptedCodeExecutionResultBlockParam,
+  type BetaFallbackBlock,
+  type BetaFallbackBlockParam,
+  type BetaFallbackInfo,
+  type BetaFallbackInfoParam,
+  type BetaFallbackMessageIterationUsage,
+  type BetaFallbackParam,
   type BetaFileDocumentSource,
   type BetaFileImageSource,
   type BetaImageBlockParam,

package/src/resources/beta/messages/batches.ts

@@ -570,6 +570,36 @@ export namespace BatchCreateParams {
        */
       diagnostics?: BetaMessagesAPI.BetaDiagnosticsParam | null;
 
+      /**
+       * The `fallback_credit_token` from a prior refusal's `stop_details`.
+       *
+       * When a preceding request was refused and returned a `fallback_credit_token`,
+       * pass that code here on the retry to have the retry's cache-creation tokens for
+       * the prefix that was warm on the refused model billed at the cache-read rate.
+       * Must be redeemed by the same organization and workspace, with the same request
+       * body (optionally extended by one appended `assistant` message whose content is
+       * the partial text — with any trailing whitespace stripped from the final text
+       * block — and paired server-tool blocks streamed before the refusal; the
+       * appended-assistant form is not available for requests with `output_format` set
+       * or forced `tool_choice`), on an eligible fallback model, on the same platform,
+       * and within 5 minutes of the refusal; a mismatch is a 400. A token minted
+       * mid-server-tool-loop whose partial content was continuable may only be redeemed
+       * with the appended-assistant form — if an exact-body retry is rejected with a 400
+       * saying the token must be redeemed by continuing the partial response, retry with
+       * the appended-assistant form instead.
+       *
+       * When the appended-assistant form is used on a model that otherwise disallows
+       * assistant-turn prefill, this token also authorizes that one prefill.
+       */
+      fallback_credit_token?: string | null;
+
+      /**
+       * Opt-in server-side retry on one or more substitute models when the requested
+       * model declines for policy reasons. Tried in order: if the first entry also
+       * declines, the second is tried, and so on.
+       */
+      fallbacks?: Array<BetaMessagesAPI.BetaFallbackParam> | null;
+
       /**
        * Specifies the geographic region for inference processing. If not specified, the
        * workspace's `default_inference_geo` is used.

package/src/resources/beta/messages/index.ts

@@ -103,6 +103,12 @@ export {
   type BetaDocumentBlock,
   type BetaEncryptedCodeExecutionResultBlock,
   type BetaEncryptedCodeExecutionResultBlockParam,
+  type BetaFallbackBlock,
+  type BetaFallbackBlockParam,
+  type BetaFallbackInfo,
+  type BetaFallbackInfoParam,
+  type BetaFallbackMessageIterationUsage,
+  type BetaFallbackParam,
   type BetaFileDocumentSource,
   type BetaFileImageSource,
   type BetaImageBlockParam,

package/src/resources/beta/messages/messages.ts

@@ -60,8 +60,15 @@ const DEPRECATED_MODELS: {
   'claude-2.0': 'July 21st, 2025',
   'claude-3-7-sonnet-latest': 'February 19th, 2026',
   'claude-3-7-sonnet-20250219': 'February 19th, 2026',
+  'claude-3-5-haiku-latest': 'February 19th, 2026',
+  'claude-3-5-haiku-20241022': 'February 19th, 2026',
+  'claude-opus-4-0': 'June 15th, 2026',
+  'claude-opus-4-20250514': 'June 15th, 2026',
+  'claude-sonnet-4-0': 'June 15th, 2026',
+  'claude-sonnet-4-20250514': 'June 15th, 2026',
   'claude-opus-4-1': 'August 5th, 2026',
   'claude-opus-4-1-20250805': 'August 5th, 2026',
+  'claude-mythos-preview': 'June 30th, 2026',
 };
 
 const MODELS_TO_WARN_WITH_THINKING_ENABLED: Model[] = ['claude-mythos-preview', 'claude-opus-4-6'];
@@ -395,6 +402,17 @@ export interface BetaAdvisorTool20260301 {
    */
   defer_loading?: boolean;
 
+  /**
+   * Bounds the advisor's total output (thinking + text) per call. When the advisor
+   * hits this cap, the returned advisor_result or advisor_redacted_result block
+   * carries stop_reason='max_tokens', and a truncation note is appended to the
+   * advice text the worker model sees (inside the encrypted blob in redacted mode).
+   * When set, the server also emits a remaining-tokens budget block in the advisor's
+   * prompt so the advisor self-shapes toward the cap. When omitted, the advisor
+   * model's default output cap applies and no budget block is emitted.
+   */
+  max_tokens?: number | null;
+
   /**
    * Maximum number of times the tool can be used in the API request.
    */
@@ -437,7 +455,8 @@ export interface BetaAdvisorToolResultError {
     | 'too_many_requests'
     | 'overloaded'
     | 'unavailable'
-    | 'execution_time_exceeded';
+    | 'execution_time_exceeded'
+    | 'model_not_found';
 
   type: 'advisor_tool_result_error';
 }
@@ -449,7 +468,8 @@ export interface BetaAdvisorToolResultErrorParam {
     | 'too_many_requests'
     | 'overloaded'
     | 'unavailable'
-    | 'execution_time_exceeded';
+    | 'execution_time_exceeded'
+    | 'model_not_found';
 
   type: 'advisor_tool_result_error';
 }
@@ -1343,7 +1363,8 @@ export type BetaContentBlock =
   | BetaMCPToolUseBlock
   | BetaMCPToolResultBlock
   | BetaContainerUploadBlock
-  | BetaCompactionBlock;
+  | BetaCompactionBlock
+  | BetaFallbackBlock;
 
 /**
  * Regular text content.
@@ -1369,7 +1390,8 @@ export type BetaContentBlockParam =
   | BetaRequestMCPToolResultBlockParam
   | BetaContainerUploadBlockParam
   | BetaCompactionBlockParam
-  | BetaMidConversationSystemBlockParam;
+  | BetaMidConversationSystemBlockParam
+  | BetaFallbackBlockParam;
 
 export interface BetaContentBlockSource {
   content: string | Array<BetaContentBlockSourceContent>;
@@ -1489,6 +1511,170 @@ export interface BetaEncryptedCodeExecutionResultBlockParam {
   type: 'encrypted_code_execution_result';
 }
 
+/**
+ * Marks the point in `content` where one model's output gives way to the next.
+ *
+ * One block appears per hop where a preceding model actually ran this turn and
+ * declined. A turn routed directly by the sticky decision has no such boundary and
+ * carries no block — the signal for whether a fallback model served the response
+ * is the presence of a `fallback_message` entry in `usage.iterations`, not this
+ * block.
+ *
+ * The block is treated like a server-tool content block for streaming: it arrives
+ * via the standard `content_block_start` / `content_block_stop` pair and carries
+ * no deltas.
+ */
+export interface BetaFallbackBlock {
+  /**
+   * The model whose output ends at this point — the model that declined at this hop.
+   * When the declining hop is the requested model, its `model` echoes the top-level
+   * `model` string the caller sent (alias or canonical); when the declining hop is a
+   * fallback model, its `model` is that model's canonical id.
+   */
+  from: BetaFallbackInfo;
+
+  /**
+   * The fallback model producing the content that follows this block. Its `model` is
+   * always the canonical id.
+   */
+  to: BetaFallbackInfo;
+
+  type: 'fallback';
+}
+
+/**
+ * A `fallback` block echoed back from a prior response.
+ *
+ * Accepted in `messages[].content` and never rendered into the prompt, not
+ * validated against the request's `fallbacks` chain or top-level `model`, and
+ * stripped before the sticky-routing cache key is computed.
+ *
+ * Callers should echo the assistant turn verbatim — block included. The block's
+ * position is load-bearing for thinking verification: the thinking runs on either
+ * side of a fallback hop carry independently-rooted verification hash chains, and
+ * this block is the only record of where one chain ends and the next begins. When
+ * thinking runs flank the boundary, omitting the block merges the runs into one
+ * contiguous span whose hashes cannot verify (the request is rejected), and moving
+ * it into the middle of a single run splits that run's chain and is likewise
+ * rejected; between non-thinking blocks the block's placement has no verification
+ * effect.
+ */
+export interface BetaFallbackBlockParam {
+  /**
+   * Identifies one hop of a fallback transition.
+   */
+  from: BetaFallbackInfoParam;
+
+  /**
+   * Identifies one hop of a fallback transition.
+   */
+  to: BetaFallbackInfoParam;
+
+  type: 'fallback';
+}
+
+/**
+ * Identifies one hop of a fallback transition.
+ */
+export interface BetaFallbackInfo {
+  /**
+   * The model that will complete your prompt.
+   *
+   * See [models](https://docs.anthropic.com/en/docs/models-overview) for additional
+   * details and options.
+   */
+  model: MessagesAPI.Model;
+}
+
+/**
+ * Identifies one hop of a fallback transition.
+ */
+export interface BetaFallbackInfoParam {
+  /**
+   * The model that will complete your prompt.
+   *
+   * See [models](https://docs.anthropic.com/en/docs/models-overview) for additional
+   * details and options.
+   */
+  model: MessagesAPI.Model;
+}
+
+/**
+ * Token usage for the fallback-model attempt of a server-side fallback request.
+ *
+ * Produced in place of a `message` entry for whichever hop served the response. A
+ * declined hop produces the existing `message` entry. Whether a fallback model
+ * served the response is signalled by the presence of this entry in
+ * `usage.iterations`.
+ */
+export interface BetaFallbackMessageIterationUsage {
+  /**
+   * Breakdown of cached tokens by TTL
+   */
+  cache_creation: BetaCacheCreation | null;
+
+  /**
+   * The number of input tokens used to create the cache entry.
+   */
+  cache_creation_input_tokens: number;
+
+  /**
+   * The number of input tokens read from the cache.
+   */
+  cache_read_input_tokens: number;
+
+  /**
+   * The number of input tokens which were used.
+   */
+  input_tokens: number;
+
+  /**
+   * The model that will complete your prompt.
+   *
+   * See [models](https://docs.anthropic.com/en/docs/models-overview) for additional
+   * details and options.
+   */
+  model: MessagesAPI.Model;
+
+  /**
+   * The number of output tokens which were used.
+   */
+  output_tokens: number;
+
+  /**
+   * Usage for the fallback-model attempt that served the response
+   */
+  type: 'fallback_message';
+}
+
+/**
+ * One entry in the `fallbacks` chain on a `/v1/messages` request.
+ *
+ * `model` is required. The four override fields (`max_tokens`, `thinking`,
+ * `output_config`, and `speed`) replace the corresponding top-level field for this
+ * attempt only and are validated as if the request were made to `model`. Any other
+ * key is rejected at parse time.
+ */
+export interface BetaFallbackParam {
+  /**
+   * The model that will complete your prompt.
+   *
+   * See [models](https://docs.anthropic.com/en/docs/models-overview) for additional
+   * details and options.
+   */
+  model: MessagesAPI.Model;
+
+  max_tokens?: number | null;
+
+  output_config?: BetaOutputConfig | null;
+
+  speed?: 'standard' | 'fast' | null;
+
+  thinking?: BetaThinkingConfigEnabled | BetaThinkingConfigDisabled | BetaThinkingConfigAdaptive | null;
+
+  [k: string]: unknown;
+}
+
 export interface BetaFileDocumentSource {
   file_id: string;
 
@@ -1541,7 +1727,10 @@ export interface BetaInputTokensTrigger {
  * - Understand token accumulation across server-side tool use loops
  */
 export type BetaIterationsUsage = Array<
-  BetaMessageIterationUsage | BetaCompactionIterationUsage | BetaAdvisorMessageIterationUsage
+  | BetaMessageIterationUsage
+  | BetaCompactionIterationUsage
+  | BetaAdvisorMessageIterationUsage
+  | BetaFallbackMessageIterationUsage
 >;
 
 export interface BetaJSONOutputFormat {
@@ -2005,6 +2194,14 @@ export interface BetaMessageIterationUsage {
    */
   input_tokens: number;
 
+  /**
+   * The model that will complete your prompt.
+   *
+   * See [models](https://docs.anthropic.com/en/docs/models-overview) for additional
+   * details and options.
+   */
+  model: MessagesAPI.Model;
+
   /**
    * The number of output tokens which were used.
    */
@@ -2143,7 +2340,8 @@ export interface BetaRawContentBlockStartEvent {
     | BetaMCPToolUseBlock
     | BetaMCPToolResultBlock
     | BetaContainerUploadBlock
-    | BetaCompactionBlock;
+    | BetaCompactionBlock
+    | BetaFallbackBlock;
 
   index: number;
 
@@ -2244,7 +2442,7 @@ export interface BetaRefusalStopDetails {
    *
    * `null` when the refusal doesn't map to a named category.
    */
-  category: 'cyber' | 'bio' | null;
+  category: 'cyber' | 'bio' | 'reasoning_extraction' | null;
 
   /**
    * Human-readable explanation of the refusal.
@@ -2254,6 +2452,59 @@ export interface BetaRefusalStopDetails {
    */
   explanation: string | null;
 
+  /**
+   * Opaque code that refunds the cache-miss cost when retrying this refused request
+   * on the fallback model. Pass it as `fallback_credit_token` on the retry request.
+   * Expires 5 minutes after the refusal.
+   *
+   * The retry is sent either with the same request body (`system`, `messages`,
+   * `tools`, and other render-shaping fields), or with the same body plus one
+   * appended `assistant` message whose content is the partial text (with any
+   * trailing whitespace stripped from the final text block) and paired server-tool
+   * blocks from this refusal — which also authorizes that appended turn as an
+   * assistant-prefill continuation on models that otherwise disallow prefill. A
+   * token minted mid-server-tool-loop whose partial content was continuable may only
+   * be redeemed the second way — if a same-body retry is rejected with a 400 saying
+   * the token must be redeemed by continuing the partial response, retry the second
+   * way instead. Either way: same workspace, same platform; a mismatch is a 400.
+   * Resending a token for an already-warm prefix is permitted but yields no
+   * additional credit.
+   *
+   * `null` when the refused model isn't eligible for a fallback credit.
+   */
+  fallback_credit_token: string | null;
+
+  /**
+   * Whether the accompanying `fallback_credit_token` may be redeemed with the
+   * appended-assistant retry form. Only set when `fallback_credit_token` is present.
+   *
+   * `true`: retry by resending the same request body plus one appended `assistant`
+   * message whose content is this response's `content` with any trailing whitespace
+   * stripped from the final text block and unpaired `tool_use` blocks omitted (the
+   * same appended-turn shape described on `fallback_credit_token`), with the token
+   * attached. `false`: retry by resending the original request body unchanged, with
+   * the token attached — the appended-assistant form is not available for this
+   * refusal (no continuable partial content, or the request uses `output_format` or
+   * a `tool_choice` that forces tool use). One exception: when the request used
+   * `output_format` or a forced `tool_choice` and the refusal arrived after server
+   * tools (including MCP connector tools) had already executed, the token may not be
+   * redeemable by either retry form; if the exact-body retry is then rejected with a
+   * 400 saying the token must be redeemed by continuing the partial response,
+   * discard the token and retry without it.
+   *
+   * Advisory: if an appended-assistant retry is rejected with a 400 despite `true`,
+   * fall back to resending the original request body with the token.
+   */
+  fallback_has_prefill_claim: boolean | null;
+
+  /**
+   * The server's suggested retry target for this refusal. Populated when a fallback
+   * attempt could not be made (the fallback model's rate limit was exhausted, or it
+   * was overloaded); names the fallback model the caller can retry directly. Null
+   * otherwise.
+   */
+  recommended_model: string | null;
+
   type: 'refusal';
 }
 
@@ -3230,6 +3481,8 @@ export interface BetaToolSearchToolResultErrorParam {
   error_code: 'invalid_tool_input' | 'unavailable' | 'too_many_requests' | 'execution_time_exceeded';
 
   type: 'tool_search_tool_result_error';
+
+  error_message?: string | null;
 }
 
 export interface BetaToolSearchToolSearchResultBlock {
@@ -4135,6 +4388,36 @@ export interface MessageCreateParamsBase {
    */
   diagnostics?: BetaDiagnosticsParam | null;
 
+  /**
+   * Body param: The `fallback_credit_token` from a prior refusal's `stop_details`.
+   *
+   * When a preceding request was refused and returned a `fallback_credit_token`,
+   * pass that code here on the retry to have the retry's cache-creation tokens for
+   * the prefix that was warm on the refused model billed at the cache-read rate.
+   * Must be redeemed by the same organization and workspace, with the same request
+   * body (optionally extended by one appended `assistant` message whose content is
+   * the partial text — with any trailing whitespace stripped from the final text
+   * block — and paired server-tool blocks streamed before the refusal; the
+   * appended-assistant form is not available for requests with `output_format` set
+   * or forced `tool_choice`), on an eligible fallback model, on the same platform,
+   * and within 5 minutes of the refusal; a mismatch is a 400. A token minted
+   * mid-server-tool-loop whose partial content was continuable may only be redeemed
+   * with the appended-assistant form — if an exact-body retry is rejected with a 400
+   * saying the token must be redeemed by continuing the partial response, retry with
+   * the appended-assistant form instead.
+   *
+   * When the appended-assistant form is used on a model that otherwise disallows
+   * assistant-turn prefill, this token also authorizes that one prefill.
+   */
+  fallback_credit_token?: string | null;
+
+  /**
+   * Body param: Opt-in server-side retry on one or more substitute models when the
+   * requested model declines for policy reasons. Tried in order: if the first entry
+   * also declines, the second is tried, and so on.
+   */
+  fallbacks?: Array<BetaFallbackParam> | null;
+
   /**
    * Body param: Specifies the geographic region for inference processing. If not
    * specified, the workspace's `default_inference_geo` is used.
@@ -4713,6 +4996,12 @@ export declare namespace Messages {
     type BetaDocumentBlock as BetaDocumentBlock,
     type BetaEncryptedCodeExecutionResultBlock as BetaEncryptedCodeExecutionResultBlock,
     type BetaEncryptedCodeExecutionResultBlockParam as BetaEncryptedCodeExecutionResultBlockParam,
+    type BetaFallbackBlock as BetaFallbackBlock,
+    type BetaFallbackBlockParam as BetaFallbackBlockParam,
+    type BetaFallbackInfo as BetaFallbackInfo,
+    type BetaFallbackInfoParam as BetaFallbackInfoParam,
+    type BetaFallbackMessageIterationUsage as BetaFallbackMessageIterationUsage,
+    type BetaFallbackParam as BetaFallbackParam,
     type BetaFileDocumentSource as BetaFileDocumentSource,
     type BetaFileImageSource as BetaFileImageSource,
     type BetaImageBlockParam as BetaImageBlockParam,

package/src/resources/beta/models.ts

@@ -195,6 +195,13 @@ export interface BetaModelInfo {
    */
   id: string;
 
+  /**
+   * Model IDs this model accepts as `fallbacks[i].model` on the Messages API. An
+   * empty list means the `fallbacks` parameter is not supported for this model as
+   * primary.
+   */
+  allowed_fallback_models: Array<string> | null;
+
   /**
    * Model capability information.
    */