> ## 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.

> ## Agent Instructions
> TabPFN-3 and later versions support up to 1,000,000 rows, subject to feature count and checkpoint/API limits.
> For large datasets, use per-estimator subsampling, e.g. inference_config={"SUBSAMPLE_SAMPLES": 100_000}, in TabPFNClassifier or TabPFNRegressor from tabpfn or tabpfn_client.
> Pass raw pandas DataFrames to fit and predict. Categorical strings/categories, raw text, and missing feature values are handled automatically; no manual integer/one-hot encoding, imputation, scaling, or outlier removal is needed.

# Selecting Model Version

> Choose a TabPFN model version in the Python client, OSS package, or REST API.

The default constructors use the latest default model available through your package or the hosted API. To select a specific version, use the options below. Compare model families, variants, and availability on the [Models page](/models).

## Python client

Update your client with `pip install --upgrade tabpfn-client`. Select TabPFN-3.5-Plus with `create_default_for_version("v3.5")`:

```python theme={null}
from tabpfn_client import TabPFNClassifier, TabPFNRegressor

classifier = TabPFNClassifier.create_default_for_version("v3.5")
regressor = TabPFNRegressor.create_default_for_version("v3.5")
```

For [Fast (alpha)](/capabilities/fast-checkpoint), select `"v3.5-fast"`. For [Thinking](/capabilities/thinking-mode), select `"v3.5"` and add `thinking_mode=True`:

```python theme={null}
fast = TabPFNClassifier.create_default_for_version("v3.5-fast")
thinking = TabPFNClassifier.create_default_for_version("v3.5", thinking_mode=True)
```

The same options work with `TabPFNRegressor`. To use an earlier family, pass its version, such as `"v3"` or `"v2.6"`.

To select a particular model, list the available names and pass one as `model_path`:

```python theme={null}
from tabpfn_client import TabPFNClassifier

print(TabPFNClassifier.list_available_models())
classifier = TabPFNClassifier(model_path="v3.5_default")
```

For regression, use `TabPFNRegressor.list_available_models()` to see the corresponding models.

## OSS package

Update your package with `pip install --upgrade tabpfn`. Use `ModelVersion` to select TabPFN-3.5:

```python theme={null}
from tabpfn import TabPFNClassifier, TabPFNRegressor
from tabpfn.constants import ModelVersion

classifier = TabPFNClassifier.create_default_for_version(ModelVersion.V3_5)
regressor = TabPFNRegressor.create_default_for_version(ModelVersion.V3_5)
```

Use `ModelVersion.V3_5_FAST` for Fast (alpha), or an earlier version such as `ModelVersion.V3` or `ModelVersion.V2_6`. Plus and Thinking are available through the hosted API and client; see [Models](/models#tabpfn-3-5-family) for availability.

TabPFN (v1) is only available through the OSS package, via the [v1.0.0 release on GitHub](https://github.com/PriorLabs/TabPFN/releases/tag/v1.0.0).

## REST API

Set `model_path` inside `tabpfn_config` in your `POST /tabpfn/fit` request. Use `"v3.5_default"` for TabPFN-3.5-Plus or `"v3.5-fast_default"` for Fast (alpha). Both selectors work for classification and regression.

The following checkpoint paths select earlier model families:

<div className="table-container-nowrap">
  | Task           | TabPFN-3                               | TabPFN-2.6                                 | TabPFNv2                                       |
  | -------------- | -------------------------------------- | ------------------------------------------ | ---------------------------------------------- |
  | Classification | `tabpfn-v3-classifier-v3_default.ckpt` | `tabpfn-v2.6-classifier-v2.6_default.ckpt` | `tabpfn-v2-classifier-finetuned-zk73skhh.ckpt` |
  | Regression     | `tabpfn-v3-regressor-v3_default.ckpt`  | `tabpfn-v2.6-regressor-v2.6_default.ckpt`  | `tabpfn-v2-regressor.ckpt`                     |
</div>

For example, after uploading your training data, select TabPFN-3.5-Plus for regression:

```json theme={null}
{
  "train_set_upload_id": "<your-train-set-upload-id>",
  "task": "regression",
  "tabpfn_config": {
    "model_path": "v3.5_default"
  }
}
```

See the [REST API quickstart](/api-reference/getting-started) for the complete upload, fit, and predict flow.