> ## Documentation Index > Fetch the complete documentation index at: https://docs.priorlabs.ai/llms.txt > Use this file to discover all available pages before exploring further. # Predict (TabPFN JSON API) > **Recommended:** Use [tabpfn-client](https://github.com/PriorLabs/tabpfn-client) (`TabPFNClassifier` / `TabPFNRegressor`). It calls these routes for you. JSON body after `POST /tabpfn/prepare_test_set_upload` and file upload. Fields: `test_set_upload_id`, `fitted_train_set_id` (from `/tabpfn/fit`), `task_config` (task + tabpfn config), optional `force_refit`. ## OpenAPI ````yaml /api-reference/openapi.json post /tabpfn/predict openapi: 3.1.0 info: title: TabPFN API description: >- Prior Labs TabPFN API. **Prefer [tabpfn-client](https://github.com/PriorLabs/tabpfn-client)**. Current integration surface: **`/tabpfn/*` JSON routes** (prepare uploads, fit, predict, limits). **`/v1/*` multipart routes are deprecated.** See the [Changelog](/changelog). version: 2.0.0 contact: name: Prior Labs email: hello@priorlabs.ai servers: - url: https://api.priorlabs.ai description: Production TabPFN API (`/tabpfn/*` current; `/v1/*` deprecated) security: - BearerAuth: [] paths: /tabpfn/predict: post: tags: - Prediction summary: Predict (TabPFN JSON API) description: >- **Recommended:** Use [tabpfn-client](https://github.com/PriorLabs/tabpfn-client) (`TabPFNClassifier` / `TabPFNRegressor`). It calls these routes for you. JSON body after `POST /tabpfn/prepare_test_set_upload` and file upload. Fields: `test_set_upload_id`, `fitted_train_set_id` (from `/tabpfn/fit`), `task_config` (task + tabpfn config), optional `force_refit`. operationId: tabpfn_predict requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/PredictRequest' example: test_set_upload_id: 223e4567-e89b-12d3-a456-426614174000 fitted_train_set_id: 323e4567-e89b-12d3-a456-426614174000 task_config: task: classification tabpfn_config: model_path: auto responses: '200': description: Prediction payload + metadata. content: application/json: schema: $ref: '#/components/schemas/PredictResponse' '401': description: >- Unauthorized — authentication required or credentials invalid. **Possible causes:** - Missing or malformed `Authorization` header - Invalid or expired JWT token - User not found (token references a deleted account) - JWT decode errors (JWEDecodeError, JWTDecodeError, JWTClaimsError) **Examples:** - Missing token: `{"detail": "Not authenticated"}` - Invalid credentials: `{"detail": "Could not validate credentials"}` content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' examples: missing_token: value: code: auth.unauthorized detail: Not authenticated retryable: false support: https://discord.com/invite/VJRuU3bSxt invalid_credentials: value: code: auth.unauthorized detail: Could not validate credentials retryable: false support: https://discord.com/invite/VJRuU3bSxt expired_token: value: code: auth.unauthorized detail: Invalid or expired JWT token retryable: false support: https://discord.com/invite/VJRuU3bSxt '404': description: >- Model not found — the provided model ID does not exist or has expired. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: code: error.not_found detail: >- Fitted model with ID 123e4567-e89b-12d3-a456-426614174000 not found retryable: false support: https://discord.com/invite/VJRuU3bSxt '422': description: Validation error — one or more fields are incorrectly formatted. content: application/json: schema: $ref: '#/components/schemas/ValidationError' '429': description: Quota exceeded. security: - BearerAuth: [] components: schemas: PredictRequest: properties: test_set_upload_id: type: string format: uuid title: Test Set Upload Id fitted_train_set_id: type: string format: uuid title: Fitted Train Set Id task_config: oneOf: - $ref: '#/components/schemas/ClassifierConfig' - $ref: '#/components/schemas/RegressorConfig' title: Task Config discriminator: propertyName: task mapping: classification: $ref: '#/components/schemas/ClassifierConfig' regression: $ref: '#/components/schemas/RegressorConfig' force_refit: type: boolean title: Force Refit description: >- Whether to force the fitting of the test set even if a fittedtest set and transform states already exist. default: false additionalProperties: false type: object required: - test_set_upload_id - fitted_train_set_id - task_config title: PredictRequest PredictResponse: properties: prediction: anyOf: - items: {} type: array - items: items: {} type: array type: array - additionalProperties: anyOf: - items: {} type: array - items: items: {} type: array type: array type: object title: Prediction metadata: oneOf: - $ref: '#/components/schemas/ClassifierMetadata' - $ref: '#/components/schemas/RegressorMetadata' title: Metadata discriminator: propertyName: task mapping: classification: $ref: '#/components/schemas/ClassifierMetadata' regression: $ref: '#/components/schemas/RegressorMetadata' additionalProperties: false type: object required: - prediction - metadata title: PredictResponse ErrorResponse: type: object required: - code - detail - retryable - support properties: code: type: string description: Error category code (e.g., auth.unauthorized, rate.limit.exceeded) example: auth.unauthorized detail: type: string description: Human-readable error message describing what went wrong. example: Invalid authentication credentials retryable: type: boolean description: Indicates whether the request can be retried. example: false support: type: string description: URL to get support for this error. example: https://discord.com/invite/VJRuU3bSxt example: code: auth.unauthorized detail: Invalid authentication credentials retryable: false support: https://discord.com/invite/VJRuU3bSxt ValidationError: type: object required: - code - detail - errors - retryable - support properties: code: type: string description: Error category code for validation errors example: validation.failed detail: type: string description: Human-readable validation error message example: Validation error for FitRequest errors: type: array description: Array of field-level validation errors items: type: object properties: field: type: string description: Field name where the error occurred example: task message: type: string description: Error message for this field example: field required retryable: type: boolean description: Indicates whether the request can be retried example: false support: type: string description: URL to get support for this error example: https://discord.com/invite/VJRuU3bSxt example: code: validation.failed detail: Validation error for FitRequest errors: - field: task message: field required - field: systems message: >- Invalid system: invalid_system. Must be one of: ['preprocessing', 'text'] retryable: false support: https://discord.com/invite/VJRuU3bSxt ClassifierConfig: properties: task: type: string const: classification title: Task default: classification tabpfn_config: $ref: '#/components/schemas/ClassifierTabPFNConfig' predict_params: $ref: '#/components/schemas/ClassifierPredictParams' additionalProperties: false type: object title: ClassifierConfig RegressorConfig: properties: task: type: string const: regression title: Task default: regression tabpfn_config: $ref: '#/components/schemas/RegressorTabPFNConfig' predict_params: $ref: '#/components/schemas/RegressorPredictParams' additionalProperties: false type: object title: RegressorConfig ClassifierMetadata: properties: test_set_num_rows: type: integer title: Test Set Num Rows test_set_num_cols: type: integer title: Test Set Num Cols task: type: string const: classification title: Task default: classification package_version: type: string title: Package Version tabpfn_config: $ref: '#/components/schemas/ClassifierTabPFNConfig' additionalProperties: false type: object required: - test_set_num_rows - test_set_num_cols - package_version - tabpfn_config title: ClassifierMetadata RegressorMetadata: properties: test_set_num_rows: type: integer title: Test Set Num Rows test_set_num_cols: type: integer title: Test Set Num Cols task: type: string const: regression title: Task default: regression package_version: type: string title: Package Version tabpfn_config: $ref: '#/components/schemas/RegressorTabPFNConfig' additionalProperties: false type: object required: - test_set_num_rows - test_set_num_cols - package_version - tabpfn_config title: RegressorMetadata ClassifierTabPFNConfig: properties: n_estimators: anyOf: - type: integer maximum: 8 minimum: 1 - type: 'null' title: N Estimators categorical_features_indices: anyOf: - items: type: integer type: array - type: 'null' title: Categorical Features Indices softmax_temperature: anyOf: - type: number - type: 'null' title: Softmax Temperature average_before_softmax: anyOf: - type: boolean - type: 'null' title: Average Before Softmax random_state: anyOf: - type: integer - type: 'null' title: Random State inference_config: anyOf: - additionalProperties: true type: object - $ref: '#/components/schemas/InferenceConfig' - type: 'null' title: Inference Config ignore_pretraining_limits: type: boolean title: Ignore Pretraining Limits default: true n_preprocessing_jobs: type: integer title: N Preprocessing Jobs default: 4 inference_precision: type: string title: Inference Precision default: auto fit_mode: $ref: '#/components/schemas/FitMode' default: fit_preprocessors device: anyOf: - items: type: string type: array minItems: 1 - type: 'null' title: Device memory_saving_mode: anyOf: - type: boolean - type: 'null' title: Memory Saving Mode model_path: anyOf: - type: string - type: 'null' title: Model Path balance_probabilities: anyOf: - type: boolean - type: 'null' title: Balance Probabilities additionalProperties: false type: object title: ClassifierTabPFNConfig ClassifierPredictParams: properties: output_type: $ref: '#/components/schemas/ClassifierOutputType' default: probas additionalProperties: false type: object title: ClassifierPredictParams RegressorTabPFNConfig: properties: n_estimators: anyOf: - type: integer maximum: 8 minimum: 1 - type: 'null' title: N Estimators categorical_features_indices: anyOf: - items: type: integer type: array - type: 'null' title: Categorical Features Indices softmax_temperature: anyOf: - type: number - type: 'null' title: Softmax Temperature average_before_softmax: anyOf: - type: boolean - type: 'null' title: Average Before Softmax random_state: anyOf: - type: integer - type: 'null' title: Random State inference_config: anyOf: - additionalProperties: true type: object - $ref: '#/components/schemas/InferenceConfig' - type: 'null' title: Inference Config ignore_pretraining_limits: type: boolean title: Ignore Pretraining Limits default: true n_preprocessing_jobs: type: integer title: N Preprocessing Jobs default: 4 inference_precision: type: string title: Inference Precision default: auto fit_mode: $ref: '#/components/schemas/FitMode' default: fit_preprocessors device: anyOf: - items: type: string type: array minItems: 1 - type: 'null' title: Device memory_saving_mode: anyOf: - type: boolean - type: 'null' title: Memory Saving Mode model_path: anyOf: - type: string - type: 'null' title: Model Path additionalProperties: false type: object title: RegressorTabPFNConfig RegressorPredictParams: properties: output_type: $ref: '#/components/schemas/RegressorOutputType' default: mean quantiles: anyOf: - items: type: number maximum: 1 minimum: 0 type: array - type: 'null' title: Quantiles additionalProperties: false type: object title: RegressorPredictParams InferenceConfig: properties: PREPROCESS_TRANSFORMS: items: $ref: '#/components/schemas/PreprocessorConfig' type: array title: Preprocess Transforms MAX_UNIQUE_FOR_CATEGORICAL_FEATURES: type: integer title: Max Unique For Categorical Features default: 30 MIN_UNIQUE_FOR_NUMERICAL_FEATURES: type: integer title: Min Unique For Numerical Features default: 4 MIN_NUMBER_SAMPLES_FOR_CATEGORICAL_INFERENCE: type: integer title: Min Number Samples For Categorical Inference default: 100 OUTLIER_REMOVAL_STD: anyOf: - type: number - type: string const: auto - type: 'null' title: Outlier Removal Std default: auto FEATURE_SHIFT_METHOD: anyOf: - type: string enum: - shuffle - rotate - type: 'null' title: Feature Shift Method default: shuffle CLASS_SHIFT_METHOD: anyOf: - type: string enum: - shuffle - rotate - type: 'null' title: Class Shift Method default: shuffle FINGERPRINT_FEATURE: type: boolean title: Fingerprint Feature default: true POLYNOMIAL_FEATURES: anyOf: - type: string enum: - 'no' - all - type: integer title: Polynomial Features default: 'no' SUBSAMPLE_SAMPLES: anyOf: - type: integer - type: number - items: {} type: array - type: 'null' title: Subsample Samples ENABLE_GPU_PREPROCESSING: type: boolean title: Enable Gpu Preprocessing default: false FEATURE_SUBSAMPLING_METHOD: type: string enum: - balanced - random - constant_and_balanced - gini_feature_importance - auto title: Feature Subsampling Method default: balanced FEATURE_SUBSAMPLING_CONSTANT_FEATURE_COUNT: type: integer title: Feature Subsampling Constant Feature Count default: 50 FEATURE_SUBSAMPLING_IMPORTANCE_TOP_K_COUNT: anyOf: - type: integer - type: number - type: string const: auto title: Feature Subsampling Importance Top K Count default: auto REGRESSION_Y_PREPROCESS_TRANSFORMS: items: anyOf: - type: string - type: 'null' type: array title: Regression Y Preprocess Transforms default: - null - safepower USE_SKLEARN_16_DECIMAL_PRECISION: type: boolean title: Use Sklearn 16 Decimal Precision default: false MAX_NUMBER_OF_CLASSES: type: integer title: Max Number Of Classes default: 10 MAX_NUMBER_OF_FEATURES: type: integer title: Max Number Of Features default: 500 MAX_NUMBER_OF_SAMPLES: type: integer title: Max Number Of Samples default: 10000 FIX_NAN_BORDERS_AFTER_TARGET_TRANSFORM: type: boolean title: Fix Nan Borders After Target Transform default: true _REGRESSION_DEFAULT_OUTLIER_REMOVAL_STD: anyOf: - type: number - type: 'null' title: Regression Default Outlier Removal Std _CLASSIFICATION_DEFAULT_OUTLIER_REMOVAL_STD: type: number title: Classification Default Outlier Removal Std default: 12 additionalProperties: false type: object required: - PREPROCESS_TRANSFORMS title: InferenceConfig description: >- Additional configuration options for inference. Several configuration options for inference are exposed in the `TabPFNClassifier` and `TabPFNRegressor` interfaces. The options in this class are more advanced and not expected to be changed by the (standard) user. Several of the preprocessing options are supported by our code for efficiency reasons (to avoid loading TabPFN multiple times). However, these can also be applied outside of the model interface. This class must be serializable as it is peristed in the model checkpoints. Do not edit the default values in this class, as this can affect the backwards compatibility of the model checkpoints. Instead, edit `get_default()`. FitMode: type: string enum: - low_memory - fit_preprocessors - fit_with_cache - batched title: FitMode ClassifierOutputType: type: string enum: - probas - preds title: ClassifierOutputType RegressorOutputType: type: string enum: - mean - median - mode - quantiles - full - main title: RegressorOutputType PreprocessorConfig: properties: name: type: string enum: - per_feature - power - safepower - power_box - safepower_box - quantile_uni_coarse - quantile_norm_coarse - quantile_uni - quantile_norm - quantile_uni_fine - quantile_norm_fine - squashing_scaler_default - squashing_scaler_max10 - robust - kdi - none - kdi_random_alpha - kdi_uni - kdi_random_alpha_uni - adaptive - norm_and_kdi - kdi_alpha_0.3_uni - kdi_alpha_0.5_uni - kdi_alpha_0.8_uni - kdi_alpha_1.0_uni - kdi_alpha_1.2_uni - kdi_alpha_1.5_uni - kdi_alpha_2.0_uni - kdi_alpha_3.0_uni - kdi_alpha_5.0_uni - kdi_alpha_0.3 - kdi_alpha_0.5 - kdi_alpha_0.8 - kdi_alpha_1.0 - kdi_alpha_1.2 - kdi_alpha_1.5 - kdi_alpha_2.0 - kdi_alpha_3.0 - kdi_alpha_5.0 title: Name categorical_name: type: string enum: - none - numeric - onehot - ordinal - ordinal_shuffled - ordinal_very_common_categories_shuffled title: Categorical Name default: none append_original: anyOf: - type: boolean - type: string const: auto title: Append Original default: false max_features_per_estimator: type: integer title: Max Features Per Estimator default: 500 global_transformer_name: anyOf: - type: string enum: - svd - svd_quarter_components - type: 'null' title: Global Transformer Name max_onehot_cardinality: anyOf: - type: integer - type: 'null' title: Max Onehot Cardinality differentiable: type: boolean title: Differentiable default: false type: object required: - name title: PreprocessorConfig description: |- Configuration for data preprocessing. Attributes: name: Name of the preprocessor. categorical_name: Name of the categorical encoding method. Options: "none", "numeric", "onehot", "ordinal", "ordinal_shuffled", "none". append_to_original: If set to "auto", this is dynamically set to True if the number of features is less than 500, and False otherwise. Note that if set to "auto" and `max_features_per_estimator` is set as well, this flag will become False if the number of features is larger than `max_features_per_estimator / 2`. If True, the transformed features are appended to the original features, however both are capped at the max_features_per_estimator threshold, this should be used with caution as a given model might not be configured for it. max_features_per_estimator: Maximum number of features per estimator. In case the dataset has more features than this, the features are subsampled for each estimator independently. If append to original is set to True we can still have more features. global_transformer_name: Name of the global transformer to use. max_onehot_cardinality: Maximum number of unique values a categorical feature can have to be one-hot encoded. Features with higher cardinality are passed through unchanged to ordinal encoding. If None, all categorical features are one-hot encoded. securitySchemes: BearerAuth: type: http scheme: bearer bearerFormat: JWT description: >- Bearer token for authentication, obtained after signing up and generating an API key. ````