Nebius Token Factory : Start Here
Table of Contents
- 0 - Sign up
- 1 - Try the UI
- 2 - Your first API call
- 3 - Building Your First Agents
- 4 - Understanding Agentic Operations and Cost
- 5 - Deep Research Agents
- 6 - Coding with Token Factory Models
- 7 - Build a fun app / demo with your coding agent
- 8 - Collect Production logs using Data Lab
- 9 - Post training your models
0 - Sign up
- Sign up at https://tokenfactory.nebius.com
- If you have credits, apply them. how to video
- Get your API key
1 - Explore models in Playground
https://tokenfactory.nebius.com/
- Inspect model cards (pricing, context .etc)
- Take a model for a spin.
- Compare two models side by side
- Checkout Artificial Analysis Index page. Compare open vs closed models.
2 - Your first API call
3 - Building Your First Agents
Simple research agent
Try simple research agent. This agent will only use built in knowledge.
Follow instructions from this project.
Tavily powered agent
In the same example, we have a research agent powered by Tavily. It will use web search in real time to fetch results.
Run them both and compare the resulting outputs.
Note: Using Tavily Agent will require TAVILY_API_KEY. You can get a free one here : https://app.tavily.com/
4 - Understanding Agentic Operations and Cost
How many turns does it take for an agent to accomplish a task? How much does it cost?
Try this example
5 - Deep Research Agents
Here are couple of examples of deep research agents
- competitive research agent - built with Tavily + Langchain + Token Factory
- pr review bot
6 - Coding with Token Factory Models
Connect your coding agent (claude, codex, opencode .etc) with models running on Token Factory.
7 - Build a fun app / demo with your coding agent
Browse Cool Apps / Demos for inspiration, then build your own.
Advanced
8 - Collect Production logs using Data Lab
Turn your production logs into training data