resources/conversations/subresources/items/methods/retrieve/index.md 2026-05-18 22:01 UTC to 2026-05-19 06:34 UTC

1## Retrieve an item

2 

3**get** `/conversations/{conversation_id}/items/{item_id}`

4 

5Get a single item from a conversation with the given IDs.

6 

7### Path Parameters

8 

9- `conversation_id: string`

10 

11- `item_id: string`

12 

13### Query Parameters

14 

15- `include: optional array of ResponseIncludable`

16 

17 Additional fields to include in the response. See the `include`

18 parameter for [listing Conversation items above](/docs/api-reference/conversations/list-items#conversations_list_items-include) for more information.

19 

20 - `"file_search_call.results"`

21 

22 - `"web_search_call.results"`

23 

24 - `"web_search_call.action.sources"`

25 

26 - `"message.input_image.image_url"`

27 

28 - `"computer_call_output.output.image_url"`

29 

30 - `"code_interpreter_call.outputs"`

31 

32 - `"reasoning.encrypted_content"`

33 

34 - `"message.output_text.logprobs"`

35 

36### Returns

37 

38- `ConversationItem = Message or object { id, arguments, call_id, 5 more } or object { id, call_id, output, 3 more } or 22 more`

39 

40 A single item within a conversation. The set of possible types are the same as the `output` type of a [Response object](/docs/api-reference/responses/object#responses/object-output).

41 

42 - `Message object { id, content, role, 3 more }`

43 

44 A message to or from the model.

45 

46 - `id: string`

47 

48 The unique ID of the message.

49 

50 - `content: array of ResponseInputText or ResponseOutputText or TextContent or 6 more`

51 

52 The content of the message

53 

54 - `ResponseInputText object { text, type }`

55 

56 A text input to the model.

57 

58 - `text: string`

59 

60 The text input to the model.

61 

62 - `type: "input_text"`

63 

64 The type of the input item. Always `input_text`.

65 

66 - `"input_text"`

67 

68 - `ResponseOutputText object { annotations, logprobs, text, type }`

69 

70 A text output from the model.

71 

72 - `annotations: array of object { file_id, filename, index, type } or object { end_index, start_index, title, 2 more } or object { container_id, end_index, file_id, 3 more } or object { file_id, index, type }`

73 

74 The annotations of the text output.

75 

76 - `FileCitation object { file_id, filename, index, type }`

77 

78 A citation to a file.

79 

80 - `file_id: string`

81 

82 The ID of the file.

83 

84 - `filename: string`

85 

86 The filename of the file cited.

87 

88 - `index: number`

89 

90 The index of the file in the list of files.

91 

92 - `type: "file_citation"`

93 

94 The type of the file citation. Always `file_citation`.

95 

96 - `"file_citation"`

97 

98 - `URLCitation object { end_index, start_index, title, 2 more }`

99 

100 A citation for a web resource used to generate a model response.

101 

102 - `end_index: number`

103 

104 The index of the last character of the URL citation in the message.

105 

106 - `start_index: number`

107 

108 The index of the first character of the URL citation in the message.

109 

110 - `title: string`

111 

112 The title of the web resource.

113 

114 - `type: "url_citation"`

115 

116 The type of the URL citation. Always `url_citation`.

117 

118 - `"url_citation"`

119 

120 - `url: string`

121 

122 The URL of the web resource.

123 

124 - `ContainerFileCitation object { container_id, end_index, file_id, 3 more }`

125 

126 A citation for a container file used to generate a model response.

127 

128 - `container_id: string`

129 

130 The ID of the container file.

131 

132 - `end_index: number`

133 

134 The index of the last character of the container file citation in the message.

135 

136 - `file_id: string`

137 

138 The ID of the file.

139 

140 - `filename: string`

141 

142 The filename of the container file cited.

143 

144 - `start_index: number`

145 

146 The index of the first character of the container file citation in the message.

147 

148 - `type: "container_file_citation"`

149 

150 The type of the container file citation. Always `container_file_citation`.

151 

152 - `"container_file_citation"`

153 

154 - `FilePath object { file_id, index, type }`

155 

156 A path to a file.

157 

158 - `file_id: string`

159 

160 The ID of the file.

161 

162 - `index: number`

163 

164 The index of the file in the list of files.

165 

166 - `type: "file_path"`

167 

168 The type of the file path. Always `file_path`.

169 

170 - `"file_path"`

171 

172 - `logprobs: array of object { token, bytes, logprob, top_logprobs }`

173 

174 - `token: string`

175 

176 - `bytes: array of number`

177 

178 - `logprob: number`

179 

180 - `top_logprobs: array of object { token, bytes, logprob }`

181 

182 - `token: string`

183 

184 - `bytes: array of number`

185 

186 - `logprob: number`

187 

188 - `text: string`

189 

190 The text output from the model.

191 

192 - `type: "output_text"`

193 

194 The type of the output text. Always `output_text`.

195 

196 - `"output_text"`

197 

198 - `TextContent object { text, type }`

199 

200 A text content.

201 

202 - `text: string`

203 

204 - `type: "text"`

205 

206 - `"text"`

207 

208 - `SummaryTextContent object { text, type }`

209 

210 A summary text from the model.

211 

212 - `text: string`

213 

214 A summary of the reasoning output from the model so far.

215 

216 - `type: "summary_text"`

217 

218 The type of the object. Always `summary_text`.

219 

220 - `"summary_text"`

221 

222 - `ReasoningText object { text, type }`

223 

224 Reasoning text from the model.

225 

226 - `text: string`

227 

228 The reasoning text from the model.

229 

230 - `type: "reasoning_text"`

231 

232 The type of the reasoning text. Always `reasoning_text`.

233 

234 - `"reasoning_text"`

235 

236 - `ResponseOutputRefusal object { refusal, type }`

237 

238 A refusal from the model.

239 

240 - `refusal: string`

241 

242 The refusal explanation from the model.

243 

244 - `type: "refusal"`

245 

246 The type of the refusal. Always `refusal`.

247 

248 - `"refusal"`

249 

250 - `ResponseInputImage object { detail, type, file_id, image_url }`

251 

252 An image input to the model. Learn about [image inputs](/docs/guides/vision).

253 

254 - `detail: "low" or "high" or "auto" or "original"`

255 

256 The detail level of the image to be sent to the model. One of `high`, `low`, `auto`, or `original`. Defaults to `auto`.

257 

258 - `"low"`

259 

260 - `"high"`

261 

262 - `"auto"`

263 

264 - `"original"`

265 

266 - `type: "input_image"`

267 

268 The type of the input item. Always `input_image`.

269 

270 - `"input_image"`

271 

272 - `file_id: optional string`

273 

274 The ID of the file to be sent to the model.

275 

276 - `image_url: optional string`

277 

278 The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.

279 

280 - `ComputerScreenshotContent object { detail, file_id, image_url, type }`

281 

282 A screenshot of a computer.

283 

284 - `detail: "low" or "high" or "auto" or "original"`

285 

286 The detail level of the screenshot image to be sent to the model. One of `high`, `low`, `auto`, or `original`. Defaults to `auto`.

287 

288 - `"low"`

289 

290 - `"high"`

291 

292 - `"auto"`

293 

294 - `"original"`

295 

296 - `file_id: string`

297 

298 The identifier of an uploaded file that contains the screenshot.

299 

300 - `image_url: string`

301 

302 The URL of the screenshot image.

303 

304 - `type: "computer_screenshot"`

305 

306 Specifies the event type. For a computer screenshot, this property is always set to `computer_screenshot`.

307 

308 - `"computer_screenshot"`

309 

310 - `ResponseInputFile object { type, detail, file_data, 3 more }`

311 

312 A file input to the model.

313 

314 - `type: "input_file"`

315 

316 The type of the input item. Always `input_file`.

317 

318 - `"input_file"`

319 

320 - `detail: optional "low" or "high"`

321 

322 The detail level of the file to be sent to the model. Use `low` for the default rendering behavior, or `high` to render the file at higher quality. Defaults to `low`.

323 

324 - `"low"`

325 

326 - `"high"`

327 

328 - `file_data: optional string`

329 

330 The content of the file to be sent to the model.

331 

332 - `file_id: optional string`

333 

334 The ID of the file to be sent to the model.

335 

336 - `file_url: optional string`

337 

338 The URL of the file to be sent to the model.

339 

340 - `filename: optional string`

341 

342 The name of the file to be sent to the model.

343 

344 - `role: "unknown" or "user" or "assistant" or 5 more`

345 

346 The role of the message. One of `unknown`, `user`, `assistant`, `system`, `critic`, `discriminator`, `developer`, or `tool`.

347 

348 - `"unknown"`

349 

350 - `"user"`

351 

352 - `"assistant"`

353 

354 - `"system"`

355 

356 - `"critic"`

357 

358 - `"discriminator"`

359 

360 - `"developer"`

361 

362 - `"tool"`

363 

364 - `status: "in_progress" or "completed" or "incomplete"`

365 

366 The status of item. One of `in_progress`, `completed`, or `incomplete`. Populated when items are returned via API.

367 

368 - `"in_progress"`

369 

370 - `"completed"`

371 

372 - `"incomplete"`

373 

374 - `type: "message"`

375 

376 The type of the message. Always set to `message`.

377 

378 - `"message"`

379 

380 - `phase: optional "commentary" or "final_answer"`

381 

382 Labels an `assistant` message as intermediate commentary (`commentary`) or the final answer (`final_answer`). For models like `gpt-5.3-codex` and beyond, when sending follow-up requests, preserve and resend phase on all assistant messages — dropping it can degrade performance. Not used for user messages.

383 

384 - `"commentary"`

385 

386 - `"final_answer"`

387 

388 - `FunctionCall object { id, arguments, call_id, 5 more }`

389 

390 - `id: string`

391 

392 The unique ID of the function tool call.

393 

394 - `arguments: string`

395 

396 A JSON string of the arguments to pass to the function.

397 

398 - `call_id: string`

399 

400 The unique ID of the function tool call generated by the model.

401 

402 - `name: string`

403 

404 The name of the function to run.

405 

406 - `status: "in_progress" or "completed" or "incomplete"`

407 

408 The status of the item. One of `in_progress`, `completed`, or

409 `incomplete`. Populated when items are returned via API.

410 

411 - `"in_progress"`

412 

413 - `"completed"`

414 

415 - `"incomplete"`

416 

417 - `type: "function_call"`

418 

419 The type of the function tool call. Always `function_call`.

420 

421 - `"function_call"`

422 

423 - `created_by: optional string`

424 

425 The identifier of the actor that created the item.

426 

427 - `namespace: optional string`

428 

429 The namespace of the function to run.

430 

431 - `FunctionCallOutput object { id, call_id, output, 3 more }`

432 

433 - `id: string`

434 

435 The unique ID of the function call tool output.

436 

437 - `call_id: string`

438 

439 The unique ID of the function tool call generated by the model.

440 

441 - `output: string or array of ResponseInputText or ResponseInputImage or ResponseInputFile`

442 

443 The output from the function call generated by your code.

444 Can be a string or an list of output content.

445 

446 - `StringOutput = string`

447 

448 A string of the output of the function call.

449 

450 - `OutputContentList = array of ResponseInputText or ResponseInputImage or ResponseInputFile`

451 

452 Text, image, or file output of the function call.

453 

454 - `ResponseInputText object { text, type }`

455 

456 A text input to the model.

457 

458 - `ResponseInputImage object { detail, type, file_id, image_url }`

459 

460 An image input to the model. Learn about [image inputs](/docs/guides/vision).

461 

462 - `ResponseInputFile object { type, detail, file_data, 3 more }`

463 

464 A file input to the model.

465 

466 - `status: "in_progress" or "completed" or "incomplete"`

467 

468 The status of the item. One of `in_progress`, `completed`, or

469 `incomplete`. Populated when items are returned via API.

470 

471 - `"in_progress"`

472 

473 - `"completed"`

474 

475 - `"incomplete"`

476 

477 - `type: "function_call_output"`

478 

479 The type of the function tool call output. Always `function_call_output`.

480 

481 - `"function_call_output"`

482 

483 - `created_by: optional string`

484 

485 The identifier of the actor that created the item.

486 

487 - `FileSearchCall object { id, queries, status, 2 more }`

488 

489 The results of a file search tool call. See the

490 [file search guide](/docs/guides/tools-file-search) for more information.

491 

492 - `id: string`

493 

494 The unique ID of the file search tool call.

495 

496 - `queries: array of string`

497 

498 The queries used to search for files.

499 

500 - `status: "in_progress" or "searching" or "completed" or 2 more`

501 

502 The status of the file search tool call. One of `in_progress`,

503 `searching`, `incomplete` or `failed`,

504 

505 - `"in_progress"`

506 

507 - `"searching"`

508 

509 - `"completed"`

510 

511 - `"incomplete"`

512 

513 - `"failed"`

514 

515 - `type: "file_search_call"`

516 

517 The type of the file search tool call. Always `file_search_call`.

518 

519 - `"file_search_call"`

520 

521 - `results: optional array of object { attributes, file_id, filename, 2 more }`

522 

523 The results of the file search tool call.

524 

525 - `attributes: optional map[string or number or boolean]`

526 

527 Set of 16 key-value pairs that can be attached to an object. This can be

528 useful for storing additional information about the object in a structured

529 format, and querying for objects via API or the dashboard. Keys are strings

530 with a maximum length of 64 characters. Values are strings with a maximum

531 length of 512 characters, booleans, or numbers.

532 

533 - `string`

534 

535 - `number`

536 

537 - `boolean`

538 

539 - `file_id: optional string`

540 

541 The unique ID of the file.

542 

543 - `filename: optional string`

544 

545 The name of the file.

546 

547 - `score: optional number`

548 

549 The relevance score of the file - a value between 0 and 1.

550 

551 - `text: optional string`

552 

553 The text that was retrieved from the file.

554 

555 - `WebSearchCall object { id, action, status, type }`

556 

557 The results of a web search tool call. See the

558 [web search guide](/docs/guides/tools-web-search) for more information.

559 

560 - `id: string`

561 

562 The unique ID of the web search tool call.

563 

564 - `action: object { query, type, queries, sources } or object { type, url } or object { pattern, type, url }`

565 

566 An object describing the specific action taken in this web search call.

567 Includes details on how the model used the web (search, open_page, find_in_page).

568 

569 - `Search object { query, type, queries, sources }`

570 

571 Action type "search" - Performs a web search query.

572 

573 - `query: string`

574 

575 [DEPRECATED] The search query.

576 

577 - `type: "search"`

578 

579 The action type.

580 

581 - `"search"`

582 

583 - `queries: optional array of string`

584 

585 The search queries.

586 

587 - `sources: optional array of object { type, url }`

588 

589 The sources used in the search.

590 

591 - `type: "url"`

592 

593 The type of source. Always `url`.

594 

595 - `"url"`

596 

597 - `url: string`

598 

599 The URL of the source.

600 

601 - `OpenPage object { type, url }`

602 

603 Action type "open_page" - Opens a specific URL from search results.

604 

605 - `type: "open_page"`

606 

607 The action type.

608 

609 - `"open_page"`

610 

611 - `url: optional string`

612 

613 The URL opened by the model.

614 

615 - `FindInPage object { pattern, type, url }`

616 

617 Action type "find_in_page": Searches for a pattern within a loaded page.

618 

619 - `pattern: string`

620 

621 The pattern or text to search for within the page.

622 

623 - `type: "find_in_page"`

624 

625 The action type.

626 

627 - `"find_in_page"`

628 

629 - `url: string`

630 

631 The URL of the page searched for the pattern.

632 

633 - `status: "in_progress" or "searching" or "completed" or "failed"`

634 

635 The status of the web search tool call.

636 

637 - `"in_progress"`

638 

639 - `"searching"`

640 

641 - `"completed"`

642 

643 - `"failed"`

644 

645 - `type: "web_search_call"`

646 

647 The type of the web search tool call. Always `web_search_call`.

648 

649 - `"web_search_call"`

650 

651 - `ImageGenerationCall object { id, result, status, type }`

652 

653 An image generation request made by the model.

654 

655 - `id: string`

656 

657 The unique ID of the image generation call.

658 

659 - `result: string`

660 

661 The generated image encoded in base64.

662 

663 - `status: "in_progress" or "completed" or "generating" or "failed"`

664 

665 The status of the image generation call.

666 

667 - `"in_progress"`

668 

669 - `"completed"`

670 

671 - `"generating"`

672 

673 - `"failed"`

674 

675 - `type: "image_generation_call"`

676 

677 The type of the image generation call. Always `image_generation_call`.

678 

679 - `"image_generation_call"`

680 

681 - `ComputerCall object { id, call_id, pending_safety_checks, 4 more }`

682 

683 A tool call to a computer use tool. See the

684 [computer use guide](/docs/guides/tools-computer-use) for more information.

685 

686 - `id: string`

687 

688 The unique ID of the computer call.

689 

690 - `call_id: string`

691 

692 An identifier used when responding to the tool call with output.

693 

694 - `pending_safety_checks: array of object { id, code, message }`

695 

696 The pending safety checks for the computer call.

697 

698 - `id: string`

699 

700 The ID of the pending safety check.

701 

702 - `code: optional string`

703 

704 The type of the pending safety check.

705 

706 - `message: optional string`

707 

708 Details about the pending safety check.

709 

710 - `status: "in_progress" or "completed" or "incomplete"`

711 

712 The status of the item. One of `in_progress`, `completed`, or

713 `incomplete`. Populated when items are returned via API.

714 

715 - `"in_progress"`

716 

717 - `"completed"`

718 

719 - `"incomplete"`

720 

721 - `type: "computer_call"`

722 

723 The type of the computer call. Always `computer_call`.

724 

725 - `"computer_call"`

726 

727 - `action: optional ComputerAction`

728 

729 A click action.

730 

731 - `Click object { button, type, x, 2 more }`

732 

733 A click action.

734 

735 - `button: "left" or "right" or "wheel" or 2 more`

736 

737 Indicates which mouse button was pressed during the click. One of `left`, `right`, `wheel`, `back`, or `forward`.

738 

739 - `"left"`

740 

741 - `"right"`

742 

743 - `"wheel"`

744 

745 - `"back"`

746 

747 - `"forward"`

748 

749 - `type: "click"`

750 

751 Specifies the event type. For a click action, this property is always `click`.

752 

753 - `"click"`

754 

755 - `x: number`

756 

757 The x-coordinate where the click occurred.

758 

759 - `y: number`

760 

761 The y-coordinate where the click occurred.

762 

763 - `keys: optional array of string`

764 

765 The keys being held while clicking.

766 

767 - `DoubleClick object { keys, type, x, y }`

768 

769 A double click action.

770 

771 - `keys: array of string`

772 

773 The keys being held while double-clicking.

774 

775 - `type: "double_click"`

776 

777 Specifies the event type. For a double click action, this property is always set to `double_click`.

778 

779 - `"double_click"`

780 

781 - `x: number`

782 

783 The x-coordinate where the double click occurred.

784 

785 - `y: number`

786 

787 The y-coordinate where the double click occurred.

788 

789 - `Drag object { path, type, keys }`

790 

791 A drag action.

792 

793 - `path: array of object { x, y }`

794 

795 An array of coordinates representing the path of the drag action. Coordinates will appear as an array of objects, eg

796 

797 ```

798 [

799 { x: 100, y: 200 },

800 { x: 200, y: 300 }

801 ]

802 ```

803 

804 - `x: number`

805 

806 The x-coordinate.

807 

808 - `y: number`

809 

810 The y-coordinate.

811 

812 - `type: "drag"`

813 

814 Specifies the event type. For a drag action, this property is always set to `drag`.

815 

816 - `"drag"`

817 

818 - `keys: optional array of string`

819 

820 The keys being held while dragging the mouse.

821 

822 - `Keypress object { keys, type }`

823 

824 A collection of keypresses the model would like to perform.

825 

826 - `keys: array of string`

827 

828 The combination of keys the model is requesting to be pressed. This is an array of strings, each representing a key.

829 

830 - `type: "keypress"`

831 

832 Specifies the event type. For a keypress action, this property is always set to `keypress`.

833 

834 - `"keypress"`

835 

836 - `Move object { type, x, y, keys }`

837 

838 A mouse move action.

839 

840 - `type: "move"`

841 

842 Specifies the event type. For a move action, this property is always set to `move`.

843 

844 - `"move"`

845 

846 - `x: number`

847 

848 The x-coordinate to move to.

849 

850 - `y: number`

851 

852 The y-coordinate to move to.

853 

854 - `keys: optional array of string`

855 

856 The keys being held while moving the mouse.

857 

858 - `Screenshot object { type }`

859 

860 A screenshot action.

861 

862 - `type: "screenshot"`

863 

864 Specifies the event type. For a screenshot action, this property is always set to `screenshot`.

865 

866 - `"screenshot"`

867 

868 - `Scroll object { scroll_x, scroll_y, type, 3 more }`

869 

870 A scroll action.

871 

872 - `scroll_x: number`

873 

874 The horizontal scroll distance.

875 

876 - `scroll_y: number`

877 

878 The vertical scroll distance.

879 

880 - `type: "scroll"`

881 

882 Specifies the event type. For a scroll action, this property is always set to `scroll`.

883 

884 - `"scroll"`

885 

886 - `x: number`

887 

888 The x-coordinate where the scroll occurred.

889 

890 - `y: number`

891 

892 The y-coordinate where the scroll occurred.

893 

894 - `keys: optional array of string`

895 

896 The keys being held while scrolling.

897 

898 - `Type object { text, type }`

899 

900 An action to type in text.

901 

902 - `text: string`

903 

904 The text to type.

905 

906 - `type: "type"`

907 

908 Specifies the event type. For a type action, this property is always set to `type`.

909 

910 - `"type"`

911 

912 - `Wait object { type }`

913 

914 A wait action.

915 

916 - `type: "wait"`

917 

918 Specifies the event type. For a wait action, this property is always set to `wait`.

919 

920 - `"wait"`

921 

922 - `actions: optional ComputerActionList`

923 

924 Flattened batched actions for `computer_use`. Each action includes an

925 `type` discriminator and action-specific fields.

926 

927 - `Click object { button, type, x, 2 more }`

928 

929 A click action.

930 

931 - `DoubleClick object { keys, type, x, y }`

932 

933 A double click action.

934 

935 - `Drag object { path, type, keys }`

936 

937 A drag action.

938 

939 - `Keypress object { keys, type }`

940 

941 A collection of keypresses the model would like to perform.

942 

943 - `Move object { type, x, y, keys }`

944 

945 A mouse move action.

946 

947 - `Screenshot object { type }`

948 

949 A screenshot action.

950 

951 - `Scroll object { scroll_x, scroll_y, type, 3 more }`

952 

953 A scroll action.

954 

955 - `Type object { text, type }`

956 

957 An action to type in text.

958 

959 - `Wait object { type }`

960 

961 A wait action.

962 

963 - `ComputerCallOutput object { id, call_id, output, 4 more }`

964 

965 - `id: string`

966 

967 The unique ID of the computer call tool output.

968 

969 - `call_id: string`

970 

971 The ID of the computer tool call that produced the output.

972 

973 - `output: ResponseComputerToolCallOutputScreenshot`

974 

975 A computer screenshot image used with the computer use tool.

976 

977 - `type: "computer_screenshot"`

978 

979 Specifies the event type. For a computer screenshot, this property is

980 always set to `computer_screenshot`.

981 

982 - `"computer_screenshot"`

983 

984 - `file_id: optional string`

985 

986 The identifier of an uploaded file that contains the screenshot.

987 

988 - `image_url: optional string`

989 

990 The URL of the screenshot image.

991 

992 - `status: "completed" or "incomplete" or "failed" or "in_progress"`

993 

994 The status of the message input. One of `in_progress`, `completed`, or

995 `incomplete`. Populated when input items are returned via API.

996 

997 - `"completed"`

998 

999 - `"incomplete"`

1000 

1001 - `"failed"`

1002 

1003 - `"in_progress"`

1004 

1005 - `type: "computer_call_output"`

1006 

1007 The type of the computer tool call output. Always `computer_call_output`.

1008 

1009 - `"computer_call_output"`

1010 

1011 - `acknowledged_safety_checks: optional array of object { id, code, message }`

1012 

1013 The safety checks reported by the API that have been acknowledged by the

1014 developer.

1015 

1016 - `id: string`

1017 

1018 The ID of the pending safety check.

1019 

1020 - `code: optional string`

1021 

1022 The type of the pending safety check.

1023 

1024 - `message: optional string`

1025 

1026 Details about the pending safety check.

1027 

1028 - `created_by: optional string`

1029 

1030 The identifier of the actor that created the item.

1031 

1032 - `ToolSearchCall object { id, arguments, call_id, 4 more }`

1033 

1034 - `id: string`

1035 

1036 The unique ID of the tool search call item.

1037 

1038 - `arguments: unknown`

1039 

1040 Arguments used for the tool search call.

1041 

1042 - `call_id: string`

1043 

1044 The unique ID of the tool search call generated by the model.

1045 

1046 - `execution: "server" or "client"`

1047 

1048 Whether tool search was executed by the server or by the client.

1049 

1050 - `"server"`

1051 

1052 - `"client"`

1053 

1054 - `status: "in_progress" or "completed" or "incomplete"`

1055 

1056 The status of the tool search call item that was recorded.

1057 

1058 - `"in_progress"`

1059 

1060 - `"completed"`

1061 

1062 - `"incomplete"`

1063 

1064 - `type: "tool_search_call"`

1065 

1066 The type of the item. Always `tool_search_call`.

1067 

1068 - `"tool_search_call"`

1069 

1070 - `created_by: optional string`

1071 

1072 The identifier of the actor that created the item.

1073 

1074 - `ToolSearchOutput object { id, call_id, execution, 4 more }`

1075 

1076 - `id: string`

1077 

1078 The unique ID of the tool search output item.

1079 

1080 - `call_id: string`

1081 

1082 The unique ID of the tool search call generated by the model.

1083 

1084 - `execution: "server" or "client"`

1085 

1086 Whether tool search was executed by the server or by the client.

1087 

1088 - `"server"`

1089 

1090 - `"client"`

1091 

1092 - `status: "in_progress" or "completed" or "incomplete"`

1093 

1094 The status of the tool search output item that was recorded.

1095 

1096 - `"in_progress"`

1097 

1098 - `"completed"`

1099 

1100 - `"incomplete"`

1101 

1102 - `tools: array of object { name, parameters, strict, 3 more } or object { type, vector_store_ids, filters, 2 more } or object { type } or 12 more`

1103 

1104 The loaded tool definitions returned by tool search.

1105 

1106 - `Function object { name, parameters, strict, 3 more }`

1107 

1108 Defines a function in your own code the model can choose to call. Learn more about [function calling](https://platform.openai.com/docs/guides/function-calling).

1109 

1110 - `name: string`

1111 

1112 The name of the function to call.

1113 

1114 - `parameters: map[unknown]`

1115 

1116 A JSON schema object describing the parameters of the function.

1117 

1118 - `strict: boolean`

1119 

1120 Whether to enforce strict parameter validation. Default `true`.

1121 

1122 - `type: "function"`

1123 

1124 The type of the function tool. Always `function`.

1125 

1126 - `"function"`

1127 

1128 - `defer_loading: optional boolean`

1129 

1130 Whether this function is deferred and loaded via tool search.

1131 

1132 - `description: optional string`

1133 

1134 A description of the function. Used by the model to determine whether or not to call the function.

1135 

1136 - `FileSearch object { type, vector_store_ids, filters, 2 more }`

1137 

1138 A tool that searches for relevant content from uploaded files. Learn more about the [file search tool](https://platform.openai.com/docs/guides/tools-file-search).

1139 

1140 - `type: "file_search"`

1141 

1142 The type of the file search tool. Always `file_search`.

1143 

1144 - `"file_search"`

1145 

1146 - `vector_store_ids: array of string`

1147 

1148 The IDs of the vector stores to search.

1149 

1150 - `filters: optional ComparisonFilter or CompoundFilter`

1151 

1152 A filter to apply.

1153 

1154 - `ComparisonFilter object { key, type, value }`

1155 

1156 A filter used to compare a specified attribute key to a given value using a defined comparison operation.

1157 

1158 - `key: string`

1159 

1160 The key to compare against the value.

1161 

1162 - `type: "eq" or "ne" or "gt" or 5 more`

1163 

1164 Specifies the comparison operator: `eq`, `ne`, `gt`, `gte`, `lt`, `lte`, `in`, `nin`.

1165 

1166 - `eq`: equals

1167 - `ne`: not equal

1168 - `gt`: greater than

1169 - `gte`: greater than or equal

1170 - `lt`: less than

1171 - `lte`: less than or equal

1172 - `in`: in

1173 - `nin`: not in

1174 

1175 - `"eq"`

1176 

1177 - `"ne"`

1178 

1179 - `"gt"`

1180 

1181 - `"gte"`

1182 

1183 - `"lt"`

1184 

1185 - `"lte"`

1186 

1187 - `"in"`

1188 

1189 - `"nin"`

1190 

1191 - `value: string or number or boolean or array of string or number`

1192 

1193 The value to compare against the attribute key; supports string, number, or boolean types.

1194 

1195 - `string`

1196 

1197 - `number`

1198 

1199 - `boolean`

1200 

1201 - `array of string or number`

1202 

1203 - `string`

1204 

1205 - `number`

1206 

1207 - `CompoundFilter object { filters, type }`

1208 

1209 Combine multiple filters using `and` or `or`.

1210 

1211 - `filters: array of ComparisonFilter or unknown`

1212 

1213 Array of filters to combine. Items can be `ComparisonFilter` or `CompoundFilter`.

1214 

1215 - `ComparisonFilter object { key, type, value }`

1216 

1217 A filter used to compare a specified attribute key to a given value using a defined comparison operation.

1218 

1219 - `unknown`

1220 

1221 - `type: "and" or "or"`

1222 

1223 Type of operation: `and` or `or`.

1224 

1225 - `"and"`

1226 

1227 - `"or"`

1228 

1229 - `max_num_results: optional number`

1230 

1231 The maximum number of results to return. This number should be between 1 and 50 inclusive.

1232 

1233 - `ranking_options: optional object { hybrid_search, ranker, score_threshold }`

1234 

1235 Ranking options for search.

1236 

1237 - `hybrid_search: optional object { embedding_weight, text_weight }`

1238 

1239 Weights that control how reciprocal rank fusion balances semantic embedding matches versus sparse keyword matches when hybrid search is enabled.

1240 

1241 - `embedding_weight: number`

1242 

1243 The weight of the embedding in the reciprocal ranking fusion.

1244 

1245 - `text_weight: number`

1246 

1247 The weight of the text in the reciprocal ranking fusion.

1248 

1249 - `ranker: optional "auto" or "default-2024-11-15"`

1250 

1251 The ranker to use for the file search.

1252 

1253 - `"auto"`

1254 

1255 - `"default-2024-11-15"`

1256 

1257 - `score_threshold: optional number`

1258 

1259 The score threshold for the file search, a number between 0 and 1. Numbers closer to 1 will attempt to return only the most relevant results, but may return fewer results.

1260 

1261 - `Computer object { type }`

1262 

1263 A tool that controls a virtual computer. Learn more about the [computer tool](https://platform.openai.com/docs/guides/tools-computer-use).

1264 

1265 - `type: "computer"`

1266 

1267 The type of the computer tool. Always `computer`.

1268 

1269 - `"computer"`

1270 

1271 - `ComputerUsePreview object { display_height, display_width, environment, type }`

1272 

1273 A tool that controls a virtual computer. Learn more about the [computer tool](https://platform.openai.com/docs/guides/tools-computer-use).

1274 

1275 - `display_height: number`

1276 

1277 The height of the computer display.

1278 

1279 - `display_width: number`

1280 

1281 The width of the computer display.

1282 

1283 - `environment: "windows" or "mac" or "linux" or 2 more`

1284 

1285 The type of computer environment to control.

1286 

1287 - `"windows"`

1288 

1289 - `"mac"`

1290 

1291 - `"linux"`

1292 

1293 - `"ubuntu"`

1294 

1295 - `"browser"`

1296 

1297 - `type: "computer_use_preview"`

1298 

1299 The type of the computer use tool. Always `computer_use_preview`.

1300 

1301 - `"computer_use_preview"`

1302 

1303 - `WebSearch object { type, filters, search_context_size, user_location }`

1304 

1305 Search the Internet for sources related to the prompt. Learn more about the

1306 [web search tool](/docs/guides/tools-web-search).

1307 

1308 - `type: "web_search" or "web_search_2025_08_26"`

1309 

1310 The type of the web search tool. One of `web_search` or `web_search_2025_08_26`.

1311 

1312 - `"web_search"`

1313 

1314 - `"web_search_2025_08_26"`

1315 

1316 - `filters: optional object { allowed_domains }`

1317 

1318 Filters for the search.

1319 

1320 - `allowed_domains: optional array of string`

1321 

1322 Allowed domains for the search. If not provided, all domains are allowed.

1323 Subdomains of the provided domains are allowed as well.

1324 

1325 Example: `["pubmed.ncbi.nlm.nih.gov"]`

1326 

1327 - `search_context_size: optional "low" or "medium" or "high"`

1328 

1329 High level guidance for the amount of context window space to use for the search. One of `low`, `medium`, or `high`. `medium` is the default.

1330 

1331 - `"low"`

1332 

1333 - `"medium"`

1334 

1335 - `"high"`

1336 

1337 - `user_location: optional object { city, country, region, 2 more }`

1338 

1339 The approximate location of the user.

1340 

1341 - `city: optional string`

1342 

1343 Free text input for the city of the user, e.g. `San Francisco`.

1344 

1345 - `country: optional string`

1346 

1347 The two-letter [ISO country code](https://en.wikipedia.org/wiki/ISO_3166-1) of the user, e.g. `US`.

1348 

1349 - `region: optional string`

1350 

1351 Free text input for the region of the user, e.g. `California`.

1352 

1353 - `timezone: optional string`

1354 

1355 The [IANA timezone](https://timeapi.io/documentation/iana-timezones) of the user, e.g. `America/Los_Angeles`.

1356 

1357 - `type: optional "approximate"`

1358 

1359 The type of location approximation. Always `approximate`.

1360 

1361 - `"approximate"`

1362 

1363 - `Mcp object { server_label, type, allowed_tools, 7 more }`

1364 

1365 Give the model access to additional tools via remote Model Context Protocol

1366 (MCP) servers. [Learn more about MCP](/docs/guides/tools-remote-mcp).

1367 

1368 - `server_label: string`

1369 

1370 A label for this MCP server, used to identify it in tool calls.

1371 

1372 - `type: "mcp"`

1373 

1374 The type of the MCP tool. Always `mcp`.

1375 

1376 - `"mcp"`

1377 

1378 - `allowed_tools: optional array of string or object { read_only, tool_names }`

1379 

1380 List of allowed tool names or a filter object.

1381 

1382 - `McpAllowedTools = array of string`

1383 

1384 A string array of allowed tool names

1385 

1386 - `McpToolFilter object { read_only, tool_names }`

1387 

1388 A filter object to specify which tools are allowed.

1389 

1390 - `read_only: optional boolean`

1391 

1392 Indicates whether or not a tool modifies data or is read-only. If an

1393 MCP server is [annotated with `readOnlyHint`](https://modelcontextprotocol.io/specification/2025-06-18/schema#toolannotations-readonlyhint),

1394 it will match this filter.

1395 

1396 - `tool_names: optional array of string`

1397 

1398 List of allowed tool names.

1399 

1400 - `authorization: optional string`

1401 

1402 An OAuth access token that can be used with a remote MCP server, either

1403 with a custom MCP server URL or a service connector. Your application

1404 must handle the OAuth authorization flow and provide the token here.

1405 

1406 - `connector_id: optional "connector_dropbox" or "connector_gmail" or "connector_googlecalendar" or 5 more`

1407 

1408 Identifier for service connectors, like those available in ChatGPT. One of

1409 `server_url` or `connector_id` must be provided. Learn more about service

1410 connectors [here](/docs/guides/tools-remote-mcp#connectors).

1411 

1412 Currently supported `connector_id` values are:

1413 

1414 - Dropbox: `connector_dropbox`

1415 - Gmail: `connector_gmail`

1416 - Google Calendar: `connector_googlecalendar`

1417 - Google Drive: `connector_googledrive`

1418 - Microsoft Teams: `connector_microsoftteams`

1419 - Outlook Calendar: `connector_outlookcalendar`

1420 - Outlook Email: `connector_outlookemail`

1421 - SharePoint: `connector_sharepoint`

1422 

1423 - `"connector_dropbox"`

1424 

1425 - `"connector_gmail"`

1426 

1427 - `"connector_googlecalendar"`

1428 

1429 - `"connector_googledrive"`

1430 

1431 - `"connector_microsoftteams"`

1432 

1433 - `"connector_outlookcalendar"`

1434 

1435 - `"connector_outlookemail"`

1436 

1437 - `"connector_sharepoint"`

1438 

1439 - `defer_loading: optional boolean`

1440 

1441 Whether this MCP tool is deferred and discovered via tool search.

1442 

1443 - `headers: optional map[string]`

1444 

1445 Optional HTTP headers to send to the MCP server. Use for authentication

1446 or other purposes.

1447 

1448 - `require_approval: optional object { always, never } or "always" or "never"`

1449 

1450 Specify which of the MCP server's tools require approval.

1451 

1452 - `McpToolApprovalFilter object { always, never }`

1453 

1454 Specify which of the MCP server's tools require approval. Can be

1455 `always`, `never`, or a filter object associated with tools

1456 that require approval.

1457 

1458 - `always: optional object { read_only, tool_names }`

1459 

1460 A filter object to specify which tools are allowed.

1461 

1462 - `read_only: optional boolean`

1463 

1464 Indicates whether or not a tool modifies data or is read-only. If an

1465 MCP server is [annotated with `readOnlyHint`](https://modelcontextprotocol.io/specification/2025-06-18/schema#toolannotations-readonlyhint),

1466 it will match this filter.

1467 

1468 - `tool_names: optional array of string`

1469 

1470 List of allowed tool names.

1471 

1472 - `never: optional object { read_only, tool_names }`

1473 

1474 A filter object to specify which tools are allowed.

1475 

1476 - `read_only: optional boolean`

1477 

1478 Indicates whether or not a tool modifies data or is read-only. If an

1479 MCP server is [annotated with `readOnlyHint`](https://modelcontextprotocol.io/specification/2025-06-18/schema#toolannotations-readonlyhint),

1480 it will match this filter.

1481 

1482 - `tool_names: optional array of string`

1483 

1484 List of allowed tool names.

1485 

1486 - `McpToolApprovalSetting = "always" or "never"`

1487 

1488 Specify a single approval policy for all tools. One of `always` or

1489 `never`. When set to `always`, all tools will require approval. When

1490 set to `never`, all tools will not require approval.

1491 

1492 - `"always"`

1493 

1494 - `"never"`

1495 

1496 - `server_description: optional string`

1497 

1498 Optional description of the MCP server, used to provide more context.

1499 

1500 - `server_url: optional string`

1501 

1502 The URL for the MCP server. One of `server_url` or `connector_id` must be

1503 provided.

1504 

1505 - `CodeInterpreter object { container, type }`

1506 

1507 A tool that runs Python code to help generate a response to a prompt.

1508 

1509 - `container: string or object { type, file_ids, memory_limit, network_policy }`

1510 

1511 The code interpreter container. Can be a container ID or an object that

1512 specifies uploaded file IDs to make available to your code, along with an

1513 optional `memory_limit` setting.

1514 

1515 - `string`

1516 

1517 The container ID.

1518 

1519 - `CodeInterpreterToolAuto object { type, file_ids, memory_limit, network_policy }`

1520 

1521 Configuration for a code interpreter container. Optionally specify the IDs of the files to run the code on.

1522 

1523 - `type: "auto"`

1524 

1525 Always `auto`.

1526 

1527 - `"auto"`

1528 

1529 - `file_ids: optional array of string`

1530 

1531 An optional list of uploaded files to make available to your code.

1532 

1533 - `memory_limit: optional "1g" or "4g" or "16g" or "64g"`

1534 

1535 The memory limit for the code interpreter container.

1536 

1537 - `"1g"`

1538 

1539 - `"4g"`

1540 

1541 - `"16g"`

1542 

1543 - `"64g"`

1544 

1545 - `network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist`

1546 

1547 Network access policy for the container.

1548 

1549 - `ContainerNetworkPolicyDisabled object { type }`

1550 

1551 - `type: "disabled"`

1552 

1553 Disable outbound network access. Always `disabled`.

1554 

1555 - `"disabled"`

1556 

1557 - `ContainerNetworkPolicyAllowlist object { allowed_domains, type, domain_secrets }`

1558 

1559 - `allowed_domains: array of string`

1560 

1561 A list of allowed domains when type is `allowlist`.

1562 

1563 - `type: "allowlist"`

1564 

1565 Allow outbound network access only to specified domains. Always `allowlist`.

1566 

1567 - `"allowlist"`

1568 

1569 - `domain_secrets: optional array of ContainerNetworkPolicyDomainSecret`

1570 

1571 Optional domain-scoped secrets for allowlisted domains.

1572 

1573 - `domain: string`

1574 

1575 The domain associated with the secret.

1576 

1577 - `name: string`

1578 

1579 The name of the secret to inject for the domain.

1580 

1581 - `value: string`

1582 

1583 The secret value to inject for the domain.

1584 

1585 - `type: "code_interpreter"`

1586 

1587 The type of the code interpreter tool. Always `code_interpreter`.

1588 

1589 - `"code_interpreter"`

1590 

1591 - `ImageGeneration object { type, action, background, 9 more }`

1592 

1593 A tool that generates images using the GPT image models.

1594 

1595 - `type: "image_generation"`

1596 

1597 The type of the image generation tool. Always `image_generation`.

1598 

1599 - `"image_generation"`

1600 

1601 - `action: optional "generate" or "edit" or "auto"`

1602 

1603 Whether to generate a new image or edit an existing image. Default: `auto`.

1604 

1605 - `"generate"`

1606 

1607 - `"edit"`

1608 

1609 - `"auto"`

1610 

1611 - `background: optional "transparent" or "opaque" or "auto"`

1612 

1613 Background type for the generated image. One of `transparent`,

1614 `opaque`, or `auto`. Default: `auto`.

1615 

1616 - `"transparent"`

1617 

1618 - `"opaque"`

1619 

1620 - `"auto"`

1621 

1622 - `input_fidelity: optional "high" or "low"`

1623 

1624 Control how much effort the model will exert to match the style and features, especially facial features, of input images. This parameter is only supported for `gpt-image-1` and `gpt-image-1.5` and later models, unsupported for `gpt-image-1-mini`. Supports `high` and `low`. Defaults to `low`.

1625 

1626 - `"high"`

1627 

1628 - `"low"`

1629 

1630 - `input_image_mask: optional object { file_id, image_url }`

1631 

1632 Optional mask for inpainting. Contains `image_url`

1633 (string, optional) and `file_id` (string, optional).

1634 

1635 - `file_id: optional string`

1636 

1637 File ID for the mask image.

1638 

1639 - `image_url: optional string`

1640 

1641 Base64-encoded mask image.

1642 

1643 - `model: optional string or "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-1.5"`

1644 

1645 The image generation model to use. Default: `gpt-image-1`.

1646 

1647 - `string`

1648 

1649 - `"gpt-image-1" or "gpt-image-1-mini" or "gpt-image-1.5"`

1650 

1651 The image generation model to use. Default: `gpt-image-1`.

1652 

1653 - `"gpt-image-1"`

1654 

1655 - `"gpt-image-1-mini"`

1656 

1657 - `"gpt-image-1.5"`

1658 

1659 - `moderation: optional "auto" or "low"`

1660 

1661 Moderation level for the generated image. Default: `auto`.

1662 

1663 - `"auto"`

1664 

1665 - `"low"`

1666 

1667 - `output_compression: optional number`

1668 

1669 Compression level for the output image. Default: 100.

1670 

1671 - `output_format: optional "png" or "webp" or "jpeg"`

1672 

1673 The output format of the generated image. One of `png`, `webp`, or

1674 `jpeg`. Default: `png`.

1675 

1676 - `"png"`

1677 

1678 - `"webp"`

1679 

1680 - `"jpeg"`

1681 

1682 - `partial_images: optional number`

1683 

1684 Number of partial images to generate in streaming mode, from 0 (default value) to 3.

1685 

1686 - `quality: optional "low" or "medium" or "high" or "auto"`

1687 

1688 The quality of the generated image. One of `low`, `medium`, `high`,

1689 or `auto`. Default: `auto`.

1690 

1691 - `"low"`

1692 

1693 - `"medium"`

1694 

1695 - `"high"`

1696 

1697 - `"auto"`

1698 

1699 - `size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or "auto"`

1700 

1701 The size of the generated images. For `gpt-image-2` and `gpt-image-2-2026-04-21`, arbitrary resolutions are supported as `WIDTHxHEIGHT` strings, for example `1536x864`. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above `2560x1440` are experimental, and the maximum supported resolution is `3840x2160`. The requested size must also satisfy the model's current pixel and edge limits. The standard sizes `1024x1024`, `1536x1024`, and `1024x1536` are supported by the GPT image models; `auto` is supported for models that allow automatic sizing. For `dall-e-2`, use one of `256x256`, `512x512`, or `1024x1024`. For `dall-e-3`, use one of `1024x1024`, `1792x1024`, or `1024x1792`.

1702 

1703 - `string`

1704 

1705 - `"1024x1024" or "1024x1536" or "1536x1024" or "auto"`

1706 

1707 The size of the generated images. For `gpt-image-2` and `gpt-image-2-2026-04-21`, arbitrary resolutions are supported as `WIDTHxHEIGHT` strings, for example `1536x864`. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above `2560x1440` are experimental, and the maximum supported resolution is `3840x2160`. The requested size must also satisfy the model's current pixel and edge limits. The standard sizes `1024x1024`, `1536x1024`, and `1024x1536` are supported by the GPT image models; `auto` is supported for models that allow automatic sizing. For `dall-e-2`, use one of `256x256`, `512x512`, or `1024x1024`. For `dall-e-3`, use one of `1024x1024`, `1792x1024`, or `1024x1792`.

1708 

1709 - `"1024x1024"`

1710 

1711 - `"1024x1536"`

1712 

1713 - `"1536x1024"`

1714 

1715 - `"auto"`

1716 

1717 - `LocalShell object { type }`

1718 

1719 A tool that allows the model to execute shell commands in a local environment.

1720 

1721 - `type: "local_shell"`

1722 

1723 The type of the local shell tool. Always `local_shell`.

1724 

1725 - `"local_shell"`

1726 

1727 - `Shell object { type, environment }`

1728 

1729 A tool that allows the model to execute shell commands.

1730 

1731 - `type: "shell"`

1732 

1733 The type of the shell tool. Always `shell`.

1734 

1735 - `"shell"`

1736 

1737 - `environment: optional ContainerAuto or LocalEnvironment or ContainerReference`

1738 

1739 - `ContainerAuto object { type, file_ids, memory_limit, 2 more }`

1740 

1741 - `type: "container_auto"`

1742 

1743 Automatically creates a container for this request

1744 

1745 - `"container_auto"`

1746 

1747 - `file_ids: optional array of string`

1748 

1749 An optional list of uploaded files to make available to your code.

1750 

1751 - `memory_limit: optional "1g" or "4g" or "16g" or "64g"`

1752 

1753 The memory limit for the container.

1754 

1755 - `"1g"`

1756 

1757 - `"4g"`

1758 

1759 - `"16g"`

1760 

1761 - `"64g"`

1762 

1763 - `network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist`

1764 

1765 Network access policy for the container.

1766 

1767 - `ContainerNetworkPolicyDisabled object { type }`

1768 

1769 - `ContainerNetworkPolicyAllowlist object { allowed_domains, type, domain_secrets }`

1770 

1771 - `skills: optional array of SkillReference or InlineSkill`

1772 

1773 An optional list of skills referenced by id or inline data.

1774 

1775 - `SkillReference object { skill_id, type, version }`

1776 

1777 - `skill_id: string`

1778 

1779 The ID of the referenced skill.

1780 

1781 - `type: "skill_reference"`

1782 

1783 References a skill created with the /v1/skills endpoint.

1784 

1785 - `"skill_reference"`

1786 

1787 - `version: optional string`

1788 

1789 Optional skill version. Use a positive integer or 'latest'. Omit for default.

1790 

1791 - `InlineSkill object { description, name, source, type }`

1792 

1793 - `description: string`

1794 

1795 The description of the skill.

1796 

1797 - `name: string`

1798 

1799 The name of the skill.

1800 

1801 - `source: InlineSkillSource`

1802 

1803 Inline skill payload

1804 

1805 - `data: string`

1806 

1807 Base64-encoded skill zip bundle.

1808 

1809 - `media_type: "application/zip"`

1810 

1811 The media type of the inline skill payload. Must be `application/zip`.

1812 

1813 - `"application/zip"`

1814 

1815 - `type: "base64"`

1816 

1817 The type of the inline skill source. Must be `base64`.

1818 

1819 - `"base64"`

1820 

1821 - `type: "inline"`

1822 

1823 Defines an inline skill for this request.

1824 

1825 - `"inline"`

1826 

1827 - `LocalEnvironment object { type, skills }`

1828 

1829 - `type: "local"`

1830 

1831 Use a local computer environment.

1832 

1833 - `"local"`

1834 

1835 - `skills: optional array of LocalSkill`

1836 

1837 An optional list of skills.

1838 

1839 - `description: string`

1840 

1841 The description of the skill.

1842 

1843 - `name: string`

1844 

1845 The name of the skill.

1846 

1847 - `path: string`

1848 

1849 The path to the directory containing the skill.

1850 

1851 - `ContainerReference object { container_id, type }`

1852 

1853 - `container_id: string`

1854 

1855 The ID of the referenced container.

1856 

1857 - `type: "container_reference"`

1858 

1859 References a container created with the /v1/containers endpoint

1860 

1861 - `"container_reference"`

1862 

1863 - `Custom object { name, type, defer_loading, 2 more }`

1864 

1865 A custom tool that processes input using a specified format. Learn more about [custom tools](/docs/guides/function-calling#custom-tools)

1866 

1867 - `name: string`

1868 

1869 The name of the custom tool, used to identify it in tool calls.

1870 

1871 - `type: "custom"`

1872 

1873 The type of the custom tool. Always `custom`.

1874 

1875 - `"custom"`

1876 

1877 - `defer_loading: optional boolean`

1878 

1879 Whether this tool should be deferred and discovered via tool search.

1880 

1881 - `description: optional string`

1882 

1883 Optional description of the custom tool, used to provide more context.

1884 

1885 - `format: optional CustomToolInputFormat`

1886 

1887 The input format for the custom tool. Default is unconstrained text.

1888 

1889 - `Text object { type }`

1890 

1891 Unconstrained free-form text.

1892 

1893 - `type: "text"`

1894 

1895 Unconstrained text format. Always `text`.

1896 

1897 - `"text"`

1898 

1899 - `Grammar object { definition, syntax, type }`

1900 

1901 A grammar defined by the user.

1902 

1903 - `definition: string`

1904 

1905 The grammar definition.

1906 

1907 - `syntax: "lark" or "regex"`

1908 

1909 The syntax of the grammar definition. One of `lark` or `regex`.

1910 

1911 - `"lark"`

1912 

1913 - `"regex"`

1914 

1915 - `type: "grammar"`

1916 

1917 Grammar format. Always `grammar`.

1918 

1919 - `"grammar"`

1920 

1921 - `Namespace object { description, name, tools, type }`

1922 

1923 Groups function/custom tools under a shared namespace.

1924 

1925 - `description: string`

1926 

1927 A description of the namespace shown to the model.

1928 

1929 - `name: string`

1930 

1931 The namespace name used in tool calls (for example, `crm`).

1932 

1933 - `tools: array of object { name, type, defer_loading, 3 more } or object { name, type, defer_loading, 2 more }`

1934 

1935 The function/custom tools available inside this namespace.

1936 

1937 - `Function object { name, type, defer_loading, 3 more }`

1938 

1939 - `name: string`

1940 

1941 - `type: "function"`

1942 

1943 - `"function"`

1944 

1945 - `defer_loading: optional boolean`

1946 

1947 Whether this function should be deferred and discovered via tool search.

1948 

1949 - `description: optional string`

1950 

1951 - `parameters: optional unknown`

1952 

1953 - `strict: optional boolean`

1954 

1955 - `Custom object { name, type, defer_loading, 2 more }`

1956 

1957 A custom tool that processes input using a specified format. Learn more about [custom tools](/docs/guides/function-calling#custom-tools)

1958 

1959 - `name: string`

1960 

1961 The name of the custom tool, used to identify it in tool calls.

1962 

1963 - `type: "custom"`

1964 

1965 The type of the custom tool. Always `custom`.

1966 

1967 - `"custom"`

1968 

1969 - `defer_loading: optional boolean`

1970 

1971 Whether this tool should be deferred and discovered via tool search.

1972 

1973 - `description: optional string`

1974 

1975 Optional description of the custom tool, used to provide more context.

1976 

1977 - `format: optional CustomToolInputFormat`

1978 

1979 The input format for the custom tool. Default is unconstrained text.

1980 

1981 - `type: "namespace"`

1982 

1983 The type of the tool. Always `namespace`.

1984 

1985 - `"namespace"`

1986 

1987 - `ToolSearch object { type, description, execution, parameters }`

1988 

1989 Hosted or BYOT tool search configuration for deferred tools.

1990 

1991 - `type: "tool_search"`

1992 

1993 The type of the tool. Always `tool_search`.

1994 

1995 - `"tool_search"`

1996