> ## Documentation Index > Fetch the complete documentation index at: https://docs.vaero.co/llms.txt > Use this file to discover all available pages before exploring further. # Style transform > Use fine-tuned model to style text. ## OpenAPI ````yaml /api-reference/openapi.json post /style/transform openapi: 3.0.0 info: title: Vaero API version: 1.0.0 description: API for file uploads to Vaero platform. servers: - url: https://vaeroapi.com/v1 description: Production server security: [] paths: /style/transform: post: tags: - Styling summary: Style transform description: Use fine-tuned model to style text. requestBody: required: true content: application/json: schema: oneOf: - title: Full Mode type: object required: - mode - model - messages properties: mode: type: string enum: - full description: Use 'full' for end-to-end inference of a prompt. example: full messages: type: array items: type: object properties: role: type: string enum: - user - assistant content: type: string description: Conversation history. Required when mode is 'full'. example: - role: user content: Write a blog post about Florence. model: type: string description: ID of the fine-tuned model to use. example: ftm-custom-name-0b5ace7e-ffb2-43cf-8a4d-0af17bc2b1d1 full_mode_options: type: object description: Options specific to 'full' mode. properties: base_model: type: string description: >- The base model to use for generating the initial response that will be styled. Supported models include GPT (gpt-4o, gpt-4o-mini, gpt-4.1, gpt-5, gpt-5.2, gpt-5.4, o3, o4-mini) and Claude (claude-sonnet-4-5, claude-sonnet-4-6, claude-opus-4-5, claude-opus-4-6, claude-opus-4-7). example: claude-sonnet-4-6 default: claude-sonnet-4-6 base_model_api_key: type: string nullable: true description: >- API key for authenticating with the base model provider. If not provided, the key stored in your project's 3rd Party API Keys settings will be used. default: null base_model_service: type: string enum: - auto - azure - anthropic - openai description: >- The service provider to use for the base model. When set to 'auto', the service is inferred from the base_model name (Claude models use Anthropic, GPT/o-series models use OpenAI). default: auto example: auto base_model_endpoint: type: string nullable: true description: >- The Azure OpenAI endpoint URL. Only used when base_model_service is 'azure'. If not provided, the endpoint stored in your project's 3rd Party API Keys settings will be used. default: null base_temperature: type: number description: >- Sampling temperature for the base model, between 0 and 2. example: 0 default: 0 stream: type: boolean description: Whether to stream the response. default: false example: false include_quality: type: boolean description: Whether to include quality metrics in the response. default: false example: false include_rouge: type: boolean description: >- Whether to include rouge score in quality metrics. Only applies when include_quality is true. default: false example: false include_compare_original: type: boolean description: >- Whether to include original text for comparison to styled text. default: false example: false include_distance: type: boolean description: >- Whether to include distance metrics comparing the original and styled text. default: false example: false skip_headings: type: boolean description: When set to true, headings are not rewritten. default: false example: false skip_h1: type: boolean description: When set to true, H1 headings are not rewritten. default: false example: false - title: Rewrite Mode type: object required: - mode - model - message properties: mode: type: string enum: - rewrite description: Use 'rewrite' to rewrite a text. example: rewrite message: type: string description: >- Input text to be styled. The text can include multiple paragraphs. Required when mode is 'rewrite'. example: The quick brown fox jumped over the lazy dog model: type: string description: ID of the fine-tuned model to use. example: ftm-custom-name-0b5ace7e-ffb2-43cf-8a4d-0af17bc2b1d1 rewrite_mode_options: type: object description: Options specific to 'rewrite' mode. properties: topic: type: string description: >- The topic of the text being styled. Concise description of the topic and does not need to be a full sentence. It is used to provide context to your fine-tuned model for styling. example: AI training on GPUs and the systems involved default: null base_prompt: type: string description: >- Deprecated. Use 'topic' instead. Prompt used to generate the text being styled. It was used to provide context to your fine-tuned model for styling. example: Write a few sentences about a quick brown fox. default: null deprecated: true preceding_context: type: string description: >- Text that precedes the input text. Helps the style model maintain continuity with preceding content. example: The quick brown fox jumped over the lazy dog. default: null stream: type: boolean description: Whether to stream the response. default: false example: false include_quality: type: boolean description: Whether to include quality metrics in the response. default: false example: false include_rouge: type: boolean description: >- Whether to include rouge score in quality metrics. Only applies when include_quality is true. default: false example: false include_compare_original: type: boolean description: >- Whether to include original text for comparison to styled text. default: false example: false include_distance: type: boolean description: >- Whether to include distance metrics comparing the original and styled text. default: false example: false skip_headings: type: boolean description: When set to true, headings are not rewritten. default: false example: false skip_h1: type: boolean description: When set to true, H1 headings are not rewritten. default: false example: false examples: full_mode: summary: Full mode example value: mode: full messages: - role: user content: Write a blog post about Florence. model: ftm-custom-name-0b5ace7e-ffb2-43cf-8a4d-0af17bc2b1d1 full_mode_options: base_model: claude-sonnet-4-6 base_model_service: auto base_temperature: 0 stream: false include_quality: false include_rouge: false include_compare_original: false include_distance: false skip_headings: false skip_h1: false rewrite_mode: summary: Rewrite mode example value: mode: rewrite message: The quick brown fox jumped over the lazy dog model: ftm-custom-name-0b5ace7e-ffb2-43cf-8a4d-0af17bc2b1d1 rewrite_mode_options: topic: A quick brown fox stream: false include_quality: false include_rouge: false include_compare_original: false include_distance: false skip_headings: false skip_h1: false responses: '200': description: A chat completion object. content: application/json: schema: type: object properties: id: type: string description: ID of the chat completion. example: chatcmpl-123 object: type: string description: Type of the object, always 'chat.completion'. example: chat.completion created: type: integer description: Unix timestamp for when the completion was created. example: 1677652288 model: type: string description: Model used for the completion. example: ftm-custom-name-0b5ace7e-ffb2-43cf-8a4d-0af17bc2b1d1 choices: type: array items: type: object properties: index: type: integer description: Index of the choice. example: 0 message: type: object properties: role: type: string description: Role of the message author. example: assistant content: type: string description: The generated message content. example: >- The lazy dog was vaulted by a brown fox with the zoomies. finish_reason: type: string description: The reason why the generation stopped. example: stop usage: type: object properties: prompt_tokens: type: integer description: Number of tokens in the prompt. example: 9 completion_tokens: type: integer description: Number of tokens in the completion. example: 12 total_tokens: type: integer description: Total number of tokens used. example: 21 quality_analytics: type: object description: >- Quality analytics comparing the styled text against the input text and training data. Only present when include_quality is true. properties: rouge: type: object description: >- ROUGE similarity scores comparing styled text against the training data. properties: rouge1: type: object description: ROUGE-1 (unigram overlap) scores. properties: personalized: type: object description: >- ROUGE-1 scores for the styled text vs. training data. properties: precision: type: number example: 0.6875 recall: type: number example: 0.0006273883533907488 fmeasure: type: number example: 0.0012536326856231125 unpersonalized: type: object description: >- ROUGE-1 scores for the input text vs. training data. properties: precision: type: number example: 0.5833333333333334 recall: type: number example: 0.0003992471339759311 fmeasure: type: number example: 0.0007979481333713309 personalization_improvement: type: object description: >- Relative improvement from styling, as a fraction. properties: precision: type: number example: 0.1786 recall: type: number example: 0.5714 fmeasure: type: number example: 0.5711 rouge2: type: object description: ROUGE-2 (bigram overlap) scores. properties: personalized: type: object description: >- ROUGE-2 scores for the styled text vs. training data. properties: precision: type: number example: 0.4 recall: type: number example: 0.00034223134839151266 fmeasure: type: number example: 0.0006838775859121217 unpersonalized: type: object description: >- ROUGE-2 scores for the input text vs. training data. properties: precision: type: number example: 0.18181818181818182 recall: type: number example: 0.00011407711613050422 fmeasure: type: number example: 0.00022801117254745484 personalization_improvement: type: object description: >- Relative improvement from styling, as a fraction. properties: precision: type: number example: 1.2 recall: type: number example: 2 fmeasure: type: number example: 1.9993 rougeL: type: object description: ROUGE-L (longest common subsequence) scores. properties: personalized: type: object description: >- ROUGE-L scores for the styled text vs. training data. properties: precision: type: number example: 0.6875 recall: type: number example: 0.0006273883533907488 fmeasure: type: number example: 0.0012536326856231125 unpersonalized: type: object description: >- ROUGE-L scores for the input text vs. training data. properties: precision: type: number example: 0.5833333333333334 recall: type: number example: 0.0003992471339759311 fmeasure: type: number example: 0.0007979481333713309 personalization_improvement: type: object description: >- Relative improvement from styling, as a fraction. properties: precision: type: number example: 0.1786 recall: type: number example: 0.5714 fmeasure: type: number example: 0.5711 sentence_length_stats: type: object description: >- Statistics on sentence length (in words) across styled, input, and training data. properties: mean: type: object properties: personalized: type: number description: Mean sentence length of the styled text. example: 16 unpersonalized: type: number description: Mean sentence length of the input text. example: 12 ground_truth: type: number description: Mean sentence length of the training data. example: 15.157374100719425 personalization_improvement: type: number description: Relative improvement from styling. example: 0.7331 median: type: object properties: personalized: type: number description: Median sentence length of the styled text. example: 16 unpersonalized: type: number description: Median sentence length of the input text. example: 12 ground_truth: type: number description: Median sentence length of the training data. example: 14 personalization_improvement: type: number description: Relative improvement from styling. example: 0 std_dev: type: object properties: personalized: type: number description: >- Standard deviation of sentence lengths in the styled text. example: 0 unpersonalized: type: number description: >- Standard deviation of sentence lengths in the input text. example: 0 ground_truth: type: number description: >- Standard deviation of sentence lengths in the training data. example: 8.524586718203045 personalization_improvement: type: number description: Relative improvement from styling. example: 0 word_length_stats: type: object description: >- Statistics on word length (in characters) across styled, input, and training data. properties: mean: type: object properties: personalized: type: number description: Mean word length of the styled text. example: 4.1875 unpersonalized: type: number description: Mean word length of the input text. example: 5.25 ground_truth: type: number description: Mean word length of the training data. example: 4.468670643122896 personalization_improvement: type: number description: Relative improvement from styling. example: 0.6401 median: type: object properties: personalized: type: number description: Median word length of the styled text. example: 4 unpersonalized: type: number description: Median word length of the input text. example: 5 ground_truth: type: number description: Median word length of the training data. example: 4 personalization_improvement: type: number description: Relative improvement from styling. example: 1 std_dev: type: object properties: personalized: type: number description: >- Standard deviation of word lengths in the styled text. example: 2.3242942477233814 unpersonalized: type: number description: >- Standard deviation of word lengths in the input text. example: 2.520747243709029 ground_truth: type: number description: >- Standard deviation of word lengths in the training data. example: 2.4604427259371198 personalization_improvement: type: number description: Relative improvement from styling. example: -0.5571 readability: type: object description: >- Readability scores across styled, input, and training data. properties: flesch_reading_ease: type: object description: >- Flesch Reading Ease scores. Higher values indicate easier-to-read text. properties: unpersonalized: type: number description: Score for the input text. example: 42.38 personalized: type: number description: Score for the styled text. example: 63.7 ground_truth: type: number description: Score for the training data. example: 63.29 personalization_improvement: type: number description: Relative improvement from styling. example: 0.9804 flesch_kincaid_grade: type: object description: >- Flesch-Kincaid Grade Level scores. Represents the U.S. school grade level needed to understand the text. properties: unpersonalized: type: number description: Score for the input text. example: 10.3 personalized: type: number description: Score for the styled text. example: 8.4 ground_truth: type: number description: Score for the training data. example: 8.5 personalization_improvement: type: number description: Relative improvement from styling. example: 0.9444 function_word_frequency: type: object description: >- Function word frequency distribution similarity to the training data. properties: jensen_shannon_distance: type: object description: >- Jensen-Shannon distance between function word distributions. Lower values indicate greater similarity to the training data. properties: personalized: type: number description: Distance for the styled text. example: 0.6689444357441154 unpersonalized: type: number description: Distance for the input text. example: 0.7306543651658907 personalization_improvement: type: number description: Relative improvement from styling. example: 0.0845 pos_frequency: type: object description: >- Part-of-speech frequency distribution similarity to the training data. properties: jensen_shannon_distance: type: object description: >- Jensen-Shannon distance between part-of-speech distributions. Lower values indicate greater similarity to the training data. properties: personalized: type: number description: Distance for the styled text. example: 0.45509363085996124 unpersonalized: type: number description: Distance for the input text. example: 0.5559353944473231 personalization_improvement: type: number description: Relative improvement from styling. example: 0.1814 lexical_density: type: object description: >- Ratio of content words to total words. Measures how informationally dense the text is. properties: personalized: type: number description: Lexical density of the styled text. example: 0.5625 unpersonalized: type: number description: Lexical density of the input text. example: 0.6666666666666666 ground_truth: type: number description: Lexical density of the training data. example: 0.4917616488513553 personalization_improvement: type: number description: Relative improvement from styling. example: 0.5956 sentence_structure_frequency: type: object description: >- Distribution of sentence structure types across styled, input, and training data. properties: simple: type: object description: Frequency of simple sentences. properties: personalized: type: number example: 0 unpersonalized: type: number example: 1 ground_truth: type: number example: 0.4064748201438849 compound: type: object description: Frequency of compound sentences. properties: personalized: type: number example: 0 unpersonalized: type: number example: 0 ground_truth: type: number example: 0.030575539568345324 complex: type: object description: Frequency of complex sentences. properties: personalized: type: number example: 1 unpersonalized: type: number example: 0 ground_truth: type: number example: 0.5134892086330936 compound-complex: type: object description: Frequency of compound-complex sentences. properties: personalized: type: number example: 0 unpersonalized: type: number example: 0 ground_truth: type: number example: 0.04946043165467626 fragment/unknown: type: object description: >- Frequency of sentence fragments or unclassified structures. properties: personalized: type: number example: 0 unpersonalized: type: number example: 0 ground_truth: type: number example: 0 jensen_shannon_distance: type: object description: >- Jensen-Shannon distance between sentence structure distributions. Lower values indicate greater similarity to the training data. properties: personalized: type: number example: 0.45652139629164 unpersonalized: type: number example: 0.519924367696954 personalization_improvement: type: number description: Relative improvement from styling. example: 0.1219 sentence_voice_frequency: type: object description: >- Distribution of active vs. passive voice across styled, input, and training data. properties: active: type: object description: Frequency of active voice sentences. properties: personalized: type: number example: 1 unpersonalized: type: number example: 1 ground_truth: type: number example: 0.9055755395683454 passive: type: object description: Frequency of passive voice sentences. properties: personalized: type: number example: 0 unpersonalized: type: number example: 0 ground_truth: type: number example: 0.09442446043165467 jensen_shannon_distance: type: object description: >- Jensen-Shannon distance between voice distributions. Lower values indicate greater similarity to the training data. properties: personalized: type: number example: 0.18410656144604748 unpersonalized: type: number example: 0.18410656144604748 personalization_improvement: type: number description: Relative improvement from styling. example: 0 punctuation: type: object description: >- Punctuation usage density across styled, input, and training data. Each density value represents the proportion of that punctuation type relative to all punctuation. properties: comma_density: type: object properties: personalized: type: number example: 0 unpersonalized: type: number example: 0 ground_truth: type: number example: 0.3713057495969909 semicolon_density: type: object properties: personalized: type: number example: 0 unpersonalized: type: number example: 0 ground_truth: type: number example: 0.011821601289629231 colon_density: type: object properties: personalized: type: number example: 0 unpersonalized: type: number example: 0 ground_truth: type: number example: 0.04674905964535196 period_density: type: object properties: personalized: type: number example: 1 unpersonalized: type: number example: 1 ground_truth: type: number example: 0.5292853304674906 question_mark_density: type: object properties: personalized: type: number example: 0 unpersonalized: type: number example: 0 ground_truth: type: number example: 0.030628694250403012 exclamation_mark_density: type: object properties: personalized: type: number example: 0 unpersonalized: type: number example: 0 ground_truth: type: number example: 0.0016120365394948952 parenthetical_density: type: object properties: personalized: type: number example: 0 unpersonalized: type: number example: 0 ground_truth: type: number example: 0.008597528210639442 jensen_shannon_distance: type: object description: >- Jensen-Shannon distance between punctuation distributions. Lower values indicate greater similarity to the training data. properties: personalized: type: number example: 0.4471617217138273 unpersonalized: type: number example: 0.4471617217138273 personalization_improvement: type: number description: Relative improvement from styling. example: 0 '400': description: Bad request. Possibly due to invalid input data. content: application/json: schema: type: object properties: error: type: string description: Error message explaining the issue. example: Invalid input data. '401': description: >- Unauthorized. The request was missing authentication or contained invalid credentials. content: application/json: schema: type: object properties: error: type: string description: Error message explaining the issue. example: Invalid API key. security: - BearerAuth: [] components: securitySchemes: BearerAuth: type: http scheme: bearer ````