Novita AI

What it does

Novita AI is a developer platform. Developers send prompts, chat messages, embedding inputs, JSONL batch files, or code, and receive model completions, embeddings, rerank results, generated media, or sandbox execution output. Its OpenAI-compatible endpoints give access to hosted language models billed per token, and the same account also offers GPU instances and agent sandboxes.

A typical integration creates an API key in the console, stores it in an environment variable, and sends Bearer-authenticated requests to the chat completions endpoint. Teams with large request volumes can submit a batch job, upload the JSONL file, poll for completion, and download the output file. Agent sandboxes can run code, manage files, and move between running, paused, and killed states. Its main distinction is offering hosted models from several providers under one API alongside GPU and sandbox infrastructure.

Novita AI suits developers and teams building AI applications who want hosted model access without running their own inference hardware. Getting started requires a Novita account, an API key, and available credit. Some models require identity verification. Rate limits apply per model in requests and tokens per minute and depend on account top-up tiers, and sandbox lifetimes have default caps. Model and product pricing varies, so check the pricing page for current rates.

What it can do

• Hosted LLM chat completions, text completions and embeddings through OpenAI-compatible endpoints
• Model catalog spanning multiple providers, listed at novita.ai/models
• Batch API for asynchronous processing of JSONL request files with a 48-hour completion window
• Isolated agent sandboxes for running code, managing files and optional browser use
• On-demand GPU instances and serverless endpoints for inference or training
• Multimodal hosted APIs for image, video and audio generation

Works with

• Hugging Face (Novita models available on the platform)
• POE (Novita models available on the platform)
• OpenAI Python client and JavaScript/TypeScript SDK compatibility

Example workflows

Chat completion from an application

Input: a user prompt and chosen model ID. Steps: create an API key, set it as NOVITA_API_KEY, send a POST to /openai/v1/chat/completions with model, messages and max_tokens. Output: a JSON response whose choices[0].message.content holds the model's reply.

Batch processing of many requests

Input: a JSONL file of requests, at most 4 MiB. Steps: upload the file to /openai/v1/files, create a batch at /openai/v1/batches, poll the batch status until completed, then download the output file. Output: a file of model responses for the batch, retrievable within the documented retention window.

Before you choose

• Rate limits are set per model in requests and tokens per minute and depend on account top-up tiers; high-volume use may need limit increases.
• Some models require identity verification, and access can be refused with a 403 response until the account is verified.
• Sandboxes have a default maximum lifetime and idle timeout, so long-running work should use pause or lifecycle settings to preserve state.

Listed under

• Coding
• Sandboxes and infrastructure
• Agent sandbox
• Developers
• GPU cloud
• Inference
• LLM API
• Model hosting

Get started

Sign up, create an API key in the key management settings, and send a test chat completion request.

1. Create an account at novita.ai and generate an API key in the key management settings, copying it immediately because it is shown once.
2. Store the key in the NOVITA_API_KEY environment variable and confirm available credit on the billing page.
3. Send a test request to https://api.novita.ai/openai/v1/chat/completions with a Bearer Authorization header and a model ID from the models list.

Model API charges use model-specific input and output token rates. Sandbox and GPU infrastructure have their own deployment terms; confirm the relevant rates and usage meters for your workload.

Source: novita.ai/pricing, checked 29 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Novita AI - AI & Agent Cloud for Developers ↗novita.ai
• Pricing | Novita AI ↗novita.ai
• ↗docs.novita.ai

Company website

novita.ai: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.