API Platform

Developer quickstart

Make your first API request in minutes. Learn the basics of the thesis platform.

import { Thesis } from "@thesis/sdk";
const client = new Thesis();

const response = await client.runs.create({
  model: "thesis-1",
  input: "Train a sentiment classifier on the imdb dataset.",
});

console.log(response.output_text);

Build paths

REST API

Make direct requests to the platform for runs, datasets, models, and experiments.

Start with the API

Thesis CLI

Build local-first workflows that orchestrate runs, sync data, and ship models.

Start with the CLI

Models

View all

Start with thesis-1 for end-to-end pipelines, or choose mini and nano variants for lower-latency, lower-cost workloads.

thesis-1

Thesis-1

New

Our flagship model for end-to-end training and evaluation pipelines.

thesis-1-mini

Thesis-1 mini

A lower-latency model for fast iteration and lightweight workflows.

thesis-1-nano

Thesis-1 nano

Our smallest model, optimized for tight loops and on-device evaluation.

thesis-vision

Thesis Vision

Multimodal model for vision and structured output workloads.

Start building

Run a model

Send your first request to the platform and read the result.

Train a model

Launch and track an experiment from CLI or notebook.

Manage datasets

Pull, push, and version datasets across environments.

Authenticate

Set up API keys and configure your environment.