Training and fine-tuning tools for building AI agents

Platforms for fine-tuning models and training agents on your own tasks. Of the 247 tools on our building AI agents page, 8 fit this narrower job. By pricing: 7 open source, 1 source-available. The most-starred open repos here are Agent Lightning (18,733 stars), OpenPipe (10,802 stars), AReaL (5,839 stars). Most come from the frameworks (5), evals & observability (2), vision (1) categories.

Updated October 20, 2026

Agent Lightning

NewframeworksOpen source

Microsoft's lightweight trainer that improves existing agents with reinforcement learning through a model proxy, without rewriting the agent's tools or control flow.

Added Oct 17, 2026

OpenPipe

frameworksOpen source + cloud

Reinforcement learning tooling, including the ART library, to train better agents.

Added Oct 16, 2026

AReaL

frameworksOpen source

Reinforcement learning system for training LLM-based agents.

Added Oct 16, 2026

Kiln

Newevals & observabilitySource-available

Free desktop app and open-source library for building AI systems with evals, RAG, agents, fine-tuning and synthetic data in one workbench.

Added Oct 18, 2026

Prime Intellect

frameworksOpen source + cloud

Decentralized compute and open RL environment hub for training agentic models.

Added Oct 16, 2026

RAGEN

frameworksOpen source

Reinforcement learning framework for multi-turn LLM agents.

Added Oct 16, 2026

AgentEnv Framework

Newevals & observabilityOpen source

AgentEnv Framework is Scale AI's open-source Python SDK and CLI for building RL environments for agents: containerized worlds with realistic apps and data behind an MCP gateway, plus tasks (as DAGs) that grade any agent in any sandbox.

Added Oct 20, 2026

NVIDIA TAO Agent Skills

NewvisionOpen source

NVIDIA's vision-model post-training toolkit, now packaged as agent skills so Claude Code or Codex can fine-tune, optimize and export vision foundation models from plain-English prompts.

Added Oct 17, 2026

GitHub repos (8)

Repo	GitHub stars	Licence	Last push
Agent Lightning ↗	18,733	MIT	2026-10-20
OpenPipe ↗	10,802	Apache License 2.0	2026-10-20
AReaL ↗	5,839	Apache License 2.0	2026-10-21
Kiln ↗	5,195	—	2026-10-21
Prime Intellect ↗	4,713	MIT	2026-10-20
RAGEN ↗	2,824	MIT	2026-08-23
AgentEnv Framework ↗	330	—	2026-10-21
NVIDIA TAO Agent Skills ↗	95	Apache License 2.0	2026-10-19

8 repos. Stars and licence come from each repo, checked when the listing was last verified.

FAQ

What are the best AI tools for building AI agents focused on training and fine-tuning?

We list 8. The first three by GitHub stars and featured status are Agent Lightning: Microsoft's lightweight trainer that improves existing agents with reinforcement learning through a model proxy, without rewriting the agent's tools or control flow.; OpenPipe: Reinforcement learning tooling, including the ART library, to train better agents.; AReaL: Reinforcement learning system for training LLM-based agents.

Are there free or open-source options?

Yes: 7 open source, 1 source-available. Pricing is as each maker states it.

More from building AI agents

Tools are grouped by what each says it does. Missing one? Submit it free.