---
title: "Changes to Create fine-tuning job"
method: POST
path: "/fine_tuning/jobs"
---

# Changes to Create fine-tuning job

`POST /fine_tuning/jobs`

> Every recorded change to this operation, newest first.

## Timeline

Changed in 21 of 397 revisions.

- **2026-10-26** `6f5c52c2ff04`: 12 info
- **2026-10-06** `e2f8d8cbc41f`: 2 info
- **2026-10-04** `09d7b6426adc`: 6 breaking
- **2026-09-29** `13c4908e9024`: 3 breaking, 1 info
- **2026-09-28** `baee5b6ef947`: 5 breaking, 5 warning, 4 info
- **2026-09-25** `9c523485a79e`: 2 info
- **2026-09-23** `f3e92166edbe`: 6 info
- **2026-08-05** `ab0c5306e390`: 2 warning, 14 info
- **2026-07-15** `4065aec90c5d`: 2 warning, 14 info
- **2026-07-09** `e0508a503b84`: 2 warning, 14 info
- **2026-05-13** `74cbcf73838f`: 13 breaking, 1 warning, 9 info
- **2026-04-29** `5b98524f8dee`: 6 breaking
- **2026-04-10** `7c540cce6eb3`: 25 breaking, 1 warning, 47 info
- **2025-03-11** `6ab43fe7798e`: 2 info
- **2024-12-18** `8fbbc11fa933`: 1 breaking, 5 info
- **2024-10-22** `f9320ebf3471`: 5 breaking, 5 info
- **2024-09-13** `4c9e3967c707`: 1 info
- **2024-09-12** `501122aa32ad`: 1 info
- **2024-08-06** `8eca3b09b705`: 1 info
- **2024-08-06** `c36d30a94622`: 1 info
- **2024-05-01** `97c9a5f08904`: 1 info

## Changes

- **2026-10-26** `6f5c52c2ff04`: 12 info
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint/ttl`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint/ttl`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/additionalProperties/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint/ttl`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/additionalProperties/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint/ttl`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/additionalProperties/oneOf[subschema #5: LabelModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint/ttl`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/additionalProperties/oneOf[subschema #5: LabelModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint/ttl`
  - added the optional property `method/anyOf[#/components/schemas/FineTuneMethod]/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint/ttl` to the response with the `200` status
  - added the optional property `method/anyOf[#/components/schemas/FineTuneMethod]/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint/ttl` to the response with the `200` status
  - added the optional property `method/anyOf[#/components/schemas/FineTuneMethod]/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/additionalProperties/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint/ttl` to the response with the `200` status
  - added the optional property `method/anyOf[#/components/schemas/FineTuneMethod]/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/additionalProperties/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint/ttl` to the response with the `200` status
  - added the optional property `method/anyOf[#/components/schemas/FineTuneMethod]/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/additionalProperties/oneOf[subschema #5: LabelModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint/ttl` to the response with the `200` status
  - added the optional property `method/anyOf[#/components/schemas/FineTuneMethod]/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/additionalProperties/oneOf[subschema #5: LabelModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint/ttl` to the response with the `200` status
- **2026-10-06** `e2f8d8cbc41f`: 2 info
  - added the optional property `error/misalignment/review_target` to the response with the `400` status
  - added the optional property `error/misalignment/review_target` to the response with the `429` status
- **2026-10-04** `09d7b6426adc`: 6 breaking
  - the request property `integrations` became not nullable
  - the request property `integrations/items/wandb/entity` became not nullable
  - the request property `integrations/items/wandb/name` became not nullable
  - the request property `seed` became not nullable
  - the request property `suffix` became not nullable
  - the request property `validation_file` became not nullable
- **2026-09-29** `13c4908e9024`: 3 breaking, 1 info
  - removed `#/components/schemas/GraderStringCheck, #/components/schemas/GraderTextSimilarity, #/components/schemas/GraderPython, #/components/schemas/GraderScoreModel, #/components/schemas/GraderLabelModel` from the `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders` request property `oneOf` list
  - the `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders` request property type changed from no type to `object`
  - the `method/anyOf[#/components/schemas/FineTuneMethod]/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders` response's property type changed from no type to `object` for status `200`
  - removed `#/components/schemas/GraderStringCheck, #/components/schemas/GraderTextSimilarity, #/components/schemas/GraderPython, #/components/schemas/GraderScoreModel, #/components/schemas/GraderLabelModel` from the `method/anyOf[#/components/schemas/FineTuneMethod]/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders` response property `oneOf` list for the response status `200`
- **2026-09-28** `baee5b6ef947`: 5 breaking, 5 warning, 4 info
  - the response property `hyperparameters` became nullable for the status `200`
  - the response property `seed` became nullable for the status `200`
  - the response property `hyperparameters` became optional for the status `200`
  - the `method` response's property type changed from `object` to no type for status `200`
  - removed the required property `method/type` from the response with the `200` status
  - removed the optional property `method/dpo` from the response with the `200` status
  - removed the optional property `method/reinforcement` from the response with the `200` status
  - removed the optional property `method/supervised` from the response with the `200` status
  - added the new `paused` enum value to the `status` response property for the response status `200`
  - added the new `pausing` enum value to the `status` response property for the response status `200`
  - added 'else' subschema to the response body for the status `200`
  - added 'if' subschema to the response body for the status `200`
  - added `#/components/schemas/FineTuneMethod, subschema #2` to the `method` response property `anyOf` list for the response status `200`
  - added `subschema #3` to the `error` response property `anyOf` list for the response status `200`
- **2026-09-25** `9c523485a79e`: 2 info
  - added the non-success response with the status `400`
  - added the non-success response with the status `429`
- **2026-09-23** `f3e92166edbe`: 6 info
  - the `hyperparameters/learning_rate_multiplier/oneOf[subschema #2]/` request property's exclusiveMinimum was decreased from `true` to `0.00`
  - the `method/dpo/hyperparameters/beta/oneOf[subschema #2]/` request property's exclusiveMinimum was decreased from `true` to `0.00`
  - the `method/dpo/hyperparameters/learning_rate_multiplier/oneOf[subschema #2]/` request property's exclusiveMinimum was decreased from `true` to `0.00`
  - the `method/reinforcement/hyperparameters/compute_multiplier/oneOf[subschema #2]/` request property's exclusiveMinimum was decreased from `true` to `0.00`
  - the `method/reinforcement/hyperparameters/learning_rate_multiplier/oneOf[subschema #2]/` request property's exclusiveMinimum was decreased from `true` to `0.00`
  - the `method/supervised/hyperparameters/learning_rate_multiplier/oneOf[subschema #2]/` request property's exclusiveMinimum was decreased from `true` to `0.00`
- **2026-08-05** `ab0c5306e390`: 2 warning, 14 info
  - added the new `max` enum value to the `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/sampling_params/reasoning_effort/anyOf[subschema #1]/` response property for the response status `200`
  - added the new `max` enum value to the `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #4: ScoreModelGrader]/sampling_params/reasoning_effort/anyOf[subschema #1]/` response property for the response status `200`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #5: LabelModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #5: LabelModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new `max` enum value to the request property `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/sampling_params/reasoning_effort/anyOf[subschema #1]/`
  - added the new `max` enum value to the request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #4: ScoreModelGrader]/sampling_params/reasoning_effort/anyOf[subschema #1]/`
  - added the optional property `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #5: LabelModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #5: LabelModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
- **2026-07-15** `4065aec90c5d`: 2 warning, 14 info
  - added the new `max` enum value to the `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/sampling_params/reasoning_effort/anyOf[subschema #1]/` response property for the response status `200`
  - added the new `max` enum value to the `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #4: ScoreModelGrader]/sampling_params/reasoning_effort/anyOf[subschema #1]/` response property for the response status `200`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #5: LabelModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #5: LabelModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new `max` enum value to the request property `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/sampling_params/reasoning_effort/anyOf[subschema #1]/`
  - added the new `max` enum value to the request property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #4: ScoreModelGrader]/sampling_params/reasoning_effort/anyOf[subschema #1]/`
  - added the optional property `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #4: ScoreModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #5: LabelModelGrader]/input/items/content/oneOf[subschema #1: Eval content item]/oneOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/oneOf[subschema #5: MultiGrader]/graders/oneOf[subschema #5: LabelModelGrader]/input/items/content/oneOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/oneOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
- **2026-07-09** `e0508a503b84`: 2 warning, 14 info
  - added the new `max` enum value to the `method/reinforcement/grader/anyOf[subschema #4: ScoreModelGrader]/sampling_params/reasoning_effort/anyOf[subschema #1]/` response property for the response status `200`
  - added the new `max` enum value to the `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #4: ScoreModelGrader]/sampling_params/reasoning_effort/anyOf[subschema #1]/` response property for the response status `200`
  - added the new optional request property `method/reinforcement/grader/anyOf[subschema #4: ScoreModelGrader]/input/items/content/anyOf[subschema #1: Eval content item]/anyOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/anyOf[subschema #4: ScoreModelGrader]/input/items/content/anyOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/anyOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #4: ScoreModelGrader]/input/items/content/anyOf[subschema #1: Eval content item]/anyOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #4: ScoreModelGrader]/input/items/content/anyOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/anyOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #5: LabelModelGrader]/input/items/content/anyOf[subschema #1: Eval content item]/anyOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new optional request property `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #5: LabelModelGrader]/input/items/content/anyOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/anyOf[subschema #2: Input text]/prompt_cache_breakpoint`
  - added the new `max` enum value to the request property `method/reinforcement/grader/anyOf[subschema #4: ScoreModelGrader]/sampling_params/reasoning_effort/anyOf[subschema #1]/`
  - added the new `max` enum value to the request property `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #4: ScoreModelGrader]/sampling_params/reasoning_effort/anyOf[subschema #1]/`
  - added the optional property `method/reinforcement/grader/anyOf[subschema #4: ScoreModelGrader]/input/items/content/anyOf[subschema #1: Eval content item]/anyOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/anyOf[subschema #4: ScoreModelGrader]/input/items/content/anyOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/anyOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #4: ScoreModelGrader]/input/items/content/anyOf[subschema #1: Eval content item]/anyOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #4: ScoreModelGrader]/input/items/content/anyOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/anyOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #5: LabelModelGrader]/input/items/content/anyOf[subschema #1: Eval content item]/anyOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
  - added the optional property `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #5: LabelModelGrader]/input/items/content/anyOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/anyOf[subschema #2: Input text]/prompt_cache_breakpoint` to the response with the `200` status
- **2026-05-13** `74cbcf73838f`: 13 breaking, 1 warning, 9 info
  - the request property `metadata` became not nullable
  - the request property `method/type` became required
  - response property `estimated_finish` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - response property `fine_tuned_model` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - response property `finished_at` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - response property `integrations` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - response property `metadata` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - response property `trained_tokens` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - response property `validation_file` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - the `error` response's property type changed from `object` to no type for status `200`
  - removed the required property `error/code` from the response with the `200` status
  - removed the required property `error/message` from the response with the `200` status
  - removed the required property `error/param` from the response with the `200` status
  - added the new `reinforcement` enum value to the `method/type` response property for the response status `200`
  - added the new optional request property `method/reinforcement`
  - added the new `reinforcement` enum value to the request property `method/type`
  - request property `metadata` list-of-types was widened by adding types `null` to media type `application/json`
  - added the optional property `method/reinforcement` to the response with the `200` status
  - added `subschema #1, subschema #2` to the `error` response property `anyOf` list for the response status `200`
  - added `subschema #1, subschema #2` to the `hyperparameters/batch_size` response property `anyOf` list for the response status `200`
  - the response property `method/type` became required for the status `200`
  - the `batch_size` response's property default value `auto` was removed for the status `200`
  - removed `subschema #1, subschema #2` from the `hyperparameters/batch_size` response property `oneOf` list for the response status `200`
- **2026-04-29** `5b98524f8dee`: 6 breaking
  - the `method/reinforcement/grader/anyOf[subschema #4: ScoreModelGrader]/input/items/content/anyOf[subschema #1: Eval content item]/anyOf[subschema #4: Input image]/image_url` request property format changed from no format to `uri`
  - the `method/reinforcement/grader/anyOf[subschema #4: ScoreModelGrader]/input/items/content/anyOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/anyOf[subschema #4: Input image]/image_url` request property format changed from no format to `uri`
  - the `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #4: ScoreModelGrader]/input/items/content/anyOf[subschema #1: Eval content item]/anyOf[subschema #4: Input image]/image_url` request property format changed from no format to `uri`
  - the `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #4: ScoreModelGrader]/input/items/content/anyOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/anyOf[subschema #4: Input image]/image_url` request property format changed from no format to `uri`
  - the `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #5: LabelModelGrader]/input/items/content/anyOf[subschema #1: Eval content item]/anyOf[subschema #4: Input image]/image_url` request property format changed from no format to `uri`
  - the `method/reinforcement/grader/anyOf[subschema #5: MultiGrader]/graders/anyOf[subschema #5: LabelModelGrader]/input/items/content/anyOf[subschema #2: An array of Input text, Output text, Input image, and Input audio]/items/anyOf[subschema #4: Input image]/image_url` request property format changed from no format to `uri`
- **2026-04-10** `7c540cce6eb3`: 25 breaking, 1 warning, 47 info
  - removed `subschema #2` from the `model` request property `anyOf` list
  - the request property `metadata` became not nullable
  - the request property `method/type` became required
  - removed `subschema #1` from the `integrations/items/type` request property `oneOf` list
  - removed `subschema #1, subschema #2` from the `hyperparameters/batch_size` request property `oneOf` list
  - removed `subschema #1, subschema #2` from the `hyperparameters/learning_rate_multiplier` request property `oneOf` list
  - removed `subschema #1, subschema #2` from the `hyperparameters/n_epochs` request property `oneOf` list
  - removed `subschema #1, subschema #2` from the `method/dpo/hyperparameters/batch_size` request property `oneOf` list
  - removed `subschema #1, subschema #2` from the `method/dpo/hyperparameters/beta` request property `oneOf` list
  - removed `subschema #1, subschema #2` from the `method/dpo/hyperparameters/learning_rate_multiplier` request property `oneOf` list
  - removed `subschema #1, subschema #2` from the `method/dpo/hyperparameters/n_epochs` request property `oneOf` list
  - removed `subschema #1, subschema #2` from the `method/supervised/hyperparameters/batch_size` request property `oneOf` list
  - removed `subschema #1, subschema #2` from the `method/supervised/hyperparameters/learning_rate_multiplier` request property `oneOf` list
  - removed `subschema #1, subschema #2` from the `method/supervised/hyperparameters/n_epochs` request property `oneOf` list
  - response property `estimated_finish` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - response property `fine_tuned_model` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - response property `finished_at` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - response property `integrations` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - response property `metadata` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - response property `trained_tokens` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - response property `validation_file` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - the `error` response's property type changed from `object` to no type for status `200`
  - removed the required property `error/code` from the response with the `200` status
  - removed the required property `error/message` from the response with the `200` status
  - removed the required property `error/param` from the response with the `200` status
  - added the new `reinforcement` enum value to the `method/type` response property for the response status `200`
  - added the new optional request property `method/reinforcement`
  - added `subschema #1` to the `integrations/items/type` request property `anyOf` list
  - added `subschema #1, subschema #2` to the `method/dpo/hyperparameters/batch_size` request property `anyOf` list
  - added `subschema #1, subschema #2` to the `method/dpo/hyperparameters/beta` request property `anyOf` list
  - added `subschema #1, subschema #2` to the `method/dpo/hyperparameters/learning_rate_multiplier` request property `anyOf` list
  - added `subschema #1, subschema #2` to the `method/dpo/hyperparameters/n_epochs` request property `anyOf` list
  - added `subschema #1, subschema #2` to the `method/supervised/hyperparameters/batch_size` request property `anyOf` list
  - added `subschema #1, subschema #2` to the `method/supervised/hyperparameters/learning_rate_multiplier` request property `anyOf` list
  - added `subschema #1, subschema #2` to the `method/supervised/hyperparameters/n_epochs` request property `anyOf` list
  - added `subschema #1: Auto, subschema #2` to the `hyperparameters/batch_size` request property `anyOf` list
  - added `subschema #1: Auto, subschema #2` to the `hyperparameters/learning_rate_multiplier` request property `anyOf` list
  - added `subschema #1: Auto, subschema #2` to the `hyperparameters/n_epochs` request property `anyOf` list
  - added `subschema #2: Preset` to the `model` request property `anyOf` list
  - the `beta` request property default value `auto` was removed
  - …33 more
- **2025-03-11** `6ab43fe7798e`: 2 info
  - added the new optional request property `metadata`
  - added the optional property `metadata` to the response with the `200` status
- **2024-12-18** `8fbbc11fa933`: 1 breaking, 5 info
  - the response property `hyperparameters/n_epochs` became optional for the status `200`
  - added the new optional request property `method`
  - request property `hyperparameters` deprecated
  - added the optional property `hyperparameters/batch_size` to the response with the `200` status
  - added the optional property `hyperparameters/learning_rate_multiplier` to the response with the `200` status
  - added the optional property `method` to the response with the `200` status
- **2024-10-22** `f9320ebf3471`: 5 breaking, 5 info
  - removed `subschema #2` from the `model` request property `anyOf` list
  - removed `subschema #1` from the `hyperparameters/batch_size` request property `oneOf` list
  - removed `subschema #1` from the `hyperparameters/learning_rate_multiplier` request property `oneOf` list
  - removed `subschema #1` from the `hyperparameters/n_epochs` request property `oneOf` list
  - added `subschema #1: Behavior` to the `hyperparameters/n_epochs` response property `oneOf` list for the response status `200`
  - added `subschema #2: Preset` to the `model` request property `anyOf` list
  - added `subschema #1: Behavior` to the `hyperparameters/batch_size` request property `oneOf` list
  - added `subschema #1: Behavior` to the `hyperparameters/learning_rate_multiplier` request property `oneOf` list
  - added `subschema #1: Behavior` to the `hyperparameters/n_epochs` request property `oneOf` list
  - removed `subschema #1` from the `hyperparameters/n_epochs` response property `oneOf` list for the response status `200`
- **2024-09-13** `4c9e3967c707`: 1 info
  - the `suffix` request property's maxLength was increased from `40` to `64`
- **2024-09-12** `501122aa32ad`: 1 info
  - the `suffix` request property's maxLength was increased from `40` to `64`
- **2024-08-06** `8eca3b09b705`: 1 info
  - added the new `gpt-4o-mini` enum value to the request property `model/anyOf[subschema #2]/`
- **2024-08-06** `c36d30a94622`: 1 info
  - added the new `gpt-4o-mini` enum value to the request property `model/anyOf[subschema #2]/`
- **2024-05-01** `97c9a5f08904`: 1 info
  - added the optional property `estimated_finish` to the response with the `200` status

---

[Operation](https://skmtc.dev/openai/apis/openapi/docs/fine_tuning/jobs/post.md) - [API](https://skmtc.dev/openai/apis/openapi.md) - [Page](https://skmtc.dev/openai/apis/openapi/changes/fine_tuning/jobs/post)