Free AI for Open Source
A grant of 15,000,000 free AI tokens plus Pro-tier API access for maintainers of actively maintained open source projects.
100% Open-Source AI Models
Every self-hosted model on Free.ai is open-source under permissive licenses (Apache 2.0, MIT, BSD). No Llama/Meta licenses. We believe AI should be transparent, auditable, and free to use.
Chat & Language Models
| Model | Use Case | License | Source |
|---|---|---|---|
| Qwen2.5-72B | General chat, reasoning, analysis | Apache 2.0 | HuggingFace |
| Qwen2.5-7B | Fast chat, lightweight tasks | Apache 2.0 | HuggingFace |
| Mistral-7B | Fast chat, general purpose | Apache 2.0 | HuggingFace |
| Phi-3 | Efficient reasoning, math | MIT | HuggingFace |
Code Models
| Model | Use Case | License | Source |
|---|---|---|---|
| Qwen2.5-Coder-32B | Code generation, debugging | Apache 2.0 | HuggingFace |
| DeepSeek-Coder-V2-Lite | Code completion, review | MIT | HuggingFace |
Image & Video Models
| Model | Use Case | License | Source |
|---|---|---|---|
| FLUX.1-schnell | Fast image generation | Apache 2.0 | HuggingFace |
| Stable Diffusion XL | High-quality image generation | OpenRAIL++ | HuggingFace |
| Kandinsky 2.2 | Artistic image generation | Apache 2.0 | HuggingFace |
| CogVideoX-2B | Video generation from text | Apache 2.0 | HuggingFace |
| AnimateDiff | Video animation | Apache 2.0 | HuggingFace |
Speech & Audio Models
| Model | Use Case | License | Source |
|---|---|---|---|
| Kokoro | Text-to-speech | Apache 2.0 | HuggingFace |
| Piper | Text-to-speech | MIT | GitHub |
| MeloTTS | Text-to-speech | MIT | GitHub |
| Chatterbox | Text-to-speech | MIT | HuggingFace |
| OuteTTS | Text-to-speech | Apache 2.0 | HuggingFace |
| faster-whisper | Speech-to-text | MIT | GitHub |
| Vosk | Speech-to-text | Apache 2.0 | GitHub |
| ACE-Step 3.5B | Music generation | Apache 2.0 | HuggingFace |
| Mel-Band RoFormer | Music separation (vocals, instrumental) | MIT | arXiv |
| SCNet | Music separation (drums, bass, other) | Apache 2.0 | arXiv |
Translation, Vision & Other Models
| Model | Use Case | License | Source |
|---|---|---|---|
| MadLAD-400 3B | Translation (450+ languages) | Apache 2.0 | HuggingFace |
| Real-ESRGAN | Image upscaling | BSD | GitHub |
| BRIA RMBG 2.0 | Background removal | Apache 2.0 | HuggingFace |
| SAM | Image segmentation | Apache 2.0 | GitHub |
| DeOldify | Photo colorization | MIT | GitHub |
| all-MiniLM-L6-v2 | Text embeddings | Apache 2.0 | HuggingFace |
| Tesseract | OCR | Apache 2.0 | GitHub |
Open Source Maintainer Program
Maintainers of actively maintained open source projects can apply for a free grant of 15,000,000 tokens, the token allowance of a year of Pro, plus API keys at the Pro request tier (100,000 requests per month). The grant covers every tool that runs on our self-hosted models. Premium external models are not part of the grant: they use purchased tokens.
Who qualifies
- You maintain the project (owner or regular committer), and its repository is public under an OSI-approved open source license
- It is actively maintained, with commits or releases in the last 90 days
- Other people use it, shown by at least one of: 50 GitHub stars, 1,000 package downloads a month, or projects that depend on it
- One grant per Free.ai account across the open source, nonprofit and startup programs, added to an account you have already created
To apply, send the repository link and a link to the evidence of use (stars, download statistics or dependents) from the email address on your Free.ai account. Apply through the contact form
FAQ
Yes. Every model we run on our GPU servers is open-source under permissive licenses -- Apache 2.0, MIT, or BSD. We deliberately exclude models with restrictive licenses like Meta's Llama license. You can inspect, use, and modify all models freely.
Meta's Llama license includes restrictions on commercial use for large companies and requires specific attribution. We only use models with truly permissive licenses (Apache 2.0, MIT, BSD) that allow unrestricted commercial and personal use by anyone.
Every model in our table links to its source on HuggingFace or GitHub. You can download the weights directly from these repositories and run them on your own hardware.
Qwen2.5-72B is our most capable chat model for complex reasoning and analysis. Qwen2.5-7B and Mistral-7B are faster options for everyday tasks. Phi-3 is excellent for efficient reasoning and math.
We use five text-to-speech models: Kokoro (Apache 2.0), Piper (MIT), MeloTTS (MIT), Chatterbox (MIT), and OuteTTS (Apache 2.0). Each has different strengths in voice quality, speed, and language support.
FLUX.1-schnell for fast generation (Apache 2.0), Stable Diffusion XL for high-quality images (OpenRAIL++), and Kandinsky 2.2 for artistic styles (Apache 2.0). For video, we use CogVideoX-2B and AnimateDiff.
If you have developed or know of an open-source model under a permissive license that you think we should host, contact us. We evaluate models based on quality, license compatibility, and user demand.
A free grant of 15,000,000 tokens plus API keys at the Pro request tier (100,000 requests per month) for maintainers of actively maintained open source projects with evidence of use. It covers the tools that run on our self-hosted models. Premium external models are not part of the grant and use purchased tokens. Apply with a link to your repository and to the evidence of use.
We evaluate every model before deploying it based on benchmark performance, real-world output quality, inference speed, and license compliance. Only models that meet our quality bar are added to the platform.
We use MadLAD-400 3B from Google, which supports 450+ languages under an Apache 2.0 license. It handles even low-resource languages that many commercial translation services do not support.
Real-ESRGAN (BSD) for image upscaling, BRIA RMBG 2.0 (Apache 2.0) for background removal, SAM (Apache 2.0) for image segmentation, and DeOldify (MIT) for photo colorization.
No: premium external models such as GPT-5, Claude and Gemini are proprietary, carry their own licensing terms, run on purchased tokens and are not part of any comp grant. Only the self-hosted models listed on this page are open source.