Sign in

LlamaFactory

Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)

7.3 Open source76.1k
Add this badge to your site
Preview of the Reviewed by AI Review Rating badge for LlamaFactory

Run or build LlamaFactory? Paste this where your reviews or press logos go. It links to this review and always shows the current score and review date. Every tool we score 7.0 or higher can use it. It is free, and it does not change a score.

Need a launch video for LlamaFactory? We make 45 to 90 second explainers from $79. See the explainer service

LlamaFactory homepage
llamafactory.readthedocs.io · captured 2026-09-01

Our verdict

A solid, actively developed project. This assessment is derived from GitHub's own repository metrics on 2026-08-24, not from hands-on testing.

Its GitHub description reads: “Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024).” It is written in Python. It has 76,062 stars and 296 contributors, and it was pushed to within the last few weeks. The Apache-2.0 licence puts no real constraint on commercial use. The most recent tagged release is v0.9.5, from 2026-05-30. Figures come from the GitHub API on 2026-11-16 and refresh weekly. Pricing: LlamaFactory is free and open source under Apache-2.0, so there is no licence fee. Your costs are the GPU time you use for fine-tuning. Documentation is at llamafactory.readthedocs.io (read 2026-11-06).

Reviewed by AI Review Rating from GitHub API data How we score

RepositoryGitHub API, refreshed weekly

76.1kStars
9.4kForks
342Watching
303Contributors
1.2kOpen issues
Licence
Apache-2.0
Language
Python
Last push
2026-11-16
View on GitHub

Releases

  1. v0.9.5 v0.9.5: Qwen3.5/3.6, Gemma 4, Transformers v5
  2. v0.9.4 v0.9.4: Goodbye 2025
  3. v0.9.3 v0.9.3: Llama4, Gemma3, Qwen3, InternVL3, Qwen2.5-Omni
  4. v0.9.2 v0.9.2: MiniCPM-o, SwanLab, APOLLO
  5. v0.9.1 v0.9.1: Many Vision Models, Qwen2.5 Coder, Gradient Fix
  6. v0.9.0 v0.9.0: Qwen2-VL, Liger-Kernel, Adam-mini

Pros and cons

Pros

Cons

Pricingverified 6 Nov 2026

TierMonthlyIncludesSeats
Open source (Apache-2.0) Free The code is free under Apache-2.0; GPU compute for fine-tuning is your own cost (github.com/hiyouga/LlamaFactory, read 2026-11-06) 1

Changelogwhat changed, most recent first

User reviews

No one has reviewed LlamaFactory here yet. Reviews are moderated before they appear, and our own score is editorial and separate from user ratings, though once reviews land they do lift it.

Used LlamaFactory? Review it.

Sign in to rate it. One account, one review per tool, so the score means something.

Create a free account Already a member? Sign in
{{ reviewsTotal }}{{ options.labels.singularReviewCountLabel }}
{{ reviewsTotal }}{{ options.labels.pluralReviewCountLabel }}
{{ options.labels.noReviewsLabel }}
{{ options.labels.newReviewButton }}
{{ userData.canReview.message }}

LlamaFactory alternatives

Side by side

ToolOur scoreFrom Free tierBest for
LlamaFactory this page 7.3 Free Yes
🥇Hostinger Our pick 1.9 $12.19 - Visit
🥈PEFT 7.7 - Yes
🥉PyTorch Image Models (timm) 7.7 - Yes
fastai 7.4 Free Yes Learners and practitioners who want high-level PyTorch components…
DeepSpeed 7 - Yes
Hugging Face Datasets 7 - Yes

Articles on LlamaFactory

Questions

Is LlamaFactory free?

Yes. LlamaFactory is free and open source, and its own repository confirms the licence (read 2026-11-06).

How much does LlamaFactory cost?

LlamaFactory is free and open source under Apache-2.0, so there is no licence fee. Your costs are the GPU time you use for fine-tuning. Documentation is at llamafactory.readthedocs.io (read 2026-11-06).

Is LlamaFactory open source?

Yes. LlamaFactory is a public repository released under Apache-2.0. You can self-host or inspect the source directly.

How is LlamaFactory scored here?

LlamaFactory scores 7.3 out of 10. That comes from a desk review of public evidence, which is capped at 8.5 out of 10: public pricing, documentation, maintenance signals and how much we could actually verify. Affiliate relationships are excluded from scoring entirely. The full rubric is published on our methodology page.

Related