Unsloth Alternatives
Tools that fine-tune or train open language models, like Unsloth. On GitHub, Unsloth describes itself as: “Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4, MiniMax-H3, Gemma 4, FLUX and more.” Below are the 7 other open-source tools we track for the same job, ranked by live maintenance data from GitHub.
Short answer: Ranked by our health score from public GitHub data on 18 October 2026, the most actively maintained alternatives to Unsloth are TRL (100), ms-swift (94) and PEFT (93). Unsloth itself is still actively maintained (health score 100), so switching is a choice, not a necessity.
Should you switch from Unsloth?
Unsloth itself is still actively maintained (health score 100), so switching is a choice, not a necessity. Full Unsloth record
6 maintained alternatives to Unsloth, ranked
| # | Tool | Status | Health | Commits, 91 days | Latest release | Licence | Stars |
|---|---|---|---|---|---|---|---|
| 1 | TRLTrain transformer language models with reinforcement learning. | Active | 100 | 663 +111% | 15 Oct 2026 | Apache-2.0 | 19.5k |
| 2 | ms-swiftUse PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.8, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, .… | Active | 94 | 343 −13% | 13 Oct 2026 | Apache-2.0 | 15.9k |
| 3 | PEFT🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning. | Active | 93 | 215 +81% | 1 Oct 2026 | Apache-2.0 | 21.8k |
| 4 | AxolotlGo ahead and axolotl questions | Active | 93 | 209 +15% | 16 Oct 2026 | Apache-2.0 | 12.6k |
| 5 | LLaMA-FactoryUnified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024) | Slowing down | 61 | 37 −55% | 30 May 2026 | Apache-2.0 | 75.6k |
| 6 | LitGPT20+ high-performance LLMs with recipes to pretrain, finetune and deploy at scale. | Slowing down | 42 | 9 −65% | 29 Jun 2026 | Apache-2.0 | 13.7k |
Ranked by status first, then health score (0 to 100). Commits are counted on each project’s default branch; the percentage compares with the same 91 days a year earlier. Stars are bookmarks, not users.
Also in this group, but stopped or in maintenance
| Tool | Status | Health | Commits, 91 days | Latest release | Licence | Stars |
|---|---|---|---|---|---|---|
| torchtunePyTorch native post-training library | Maintenance mode | 3 | 0 −100% | 7 Apr 2025 | BSD-3-Clause | 5.8k |
These projects are archived, in maintenance mode or had no commits in the last 91 days. They may still work, but they are a risky choice for a new project.
Compare and explore
Unsloth alternatives: common questions
What is the best alternative to Unsloth?
No single tool is best for every project. By maintenance signals on 18 October 2026, the highest-ranked actively maintained alternative is TRL (health score 100, 663 commits in the last 91 days). Compare features and the licence before you switch; we have not tested these tools ourselves.
Is Unsloth still maintained?
Yes. As of 18 October 2026, Unsloth received 5,040 commits in the last 91 days and its latest commit is from 18 October 2026, and no maintenance notice was found. Its health score is 100 out of 100.
Which Unsloth alternatives have a permissive licence?
Among the actively maintained alternatives, these use a permissive licence according to GitHub: TRL (Apache-2.0), ms-swift (Apache-2.0), PEFT (Apache-2.0), Axolotl (Apache-2.0). Always read the licence file itself before commercial use.
How are these alternatives chosen and ranked?
We list the tools we track that do the same job as Unsloth (Fine-tuning and training libraries). They are ranked by maintenance status first (active, slowing down, stopped) and by our health score second. The score combines commit activity, the trend against last year, the date of the last commit, the last release and merged pull requests, all from GitHub. It does not measure features, speed, quality or security.
How this works: we group tools by the job they do and rank the ones we track with public GitHub data (commits on the default branch, releases, merged pull requests, archived flag, and maintenance notices in the README), refreshed automatically. We have not tested these tools ourselves, so the ranking says how well each project is maintained, not which one has the features you need.