Tīrtha Model Licences
Spanda Works LLC · Version 2026-10-29 · [email protected]
Every model on Tīrtha is made by another company and comes with that maker's licence. When you call a model through Tīrtha, its licence and use policy apply to your use of it, and to anyone you let use it through your product. Where a maker's policy is stricter than our Terms, the stricter rule applies. Each model's card on /models names its licence.
We list only models whose licence allows commercial use. Plain-English summaries below are for convenience; the licence text is what counts. This page is not legal advice.
What each licence asks of you
| Models | Licence | What it asks of you |
|---|---|---|
| MiniMax M3 and M3.1-Flash (MiniMax) | MiniMax's licence for that model | Commercial use requires notifying MiniMax first; Tīrtha has. Show “Built with MiniMax M3” (or “Built with MiniMax M3.1-Flash”) on a website related to a product built with it. |
| Llama 3.1, 3.3 and 4 (Meta) | Llama Community License for that version, and its Acceptable Use Policy | Follow the Acceptable Use Policy. If you build a product with a Llama model, show “Built with Llama” in it or its docs. A product with more than 700 million monthly users needs its own licence from Meta. For Llama 4 multimodal models, Meta does not grant the licence to a company whose principal place of business is in the European Union; end users of a product are not affected. Llama 3.3 licence · Llama 3.3 use policy · Llama 3.1 licence · Llama 4 licence · Llama 4 use policy |
| Gemma 3 (Google) | Gemma Terms of Use and the Gemma Prohibited Use Policy | The Prohibited Use Policy applies to you and to anyone you let use the model. If you offer a Gemma 3 model to others, you must pass the same restrictions on to them. Gemma 4 models are Apache 2.0 and carry no such policy. Gemma Terms of Use · Gemma Prohibited Use Policy |
| DeepSeek-V3, original release (DeepSeek) | DeepSeek License Agreement 1.0 | Its use restrictions (Attachment A) apply to you, and you must pass them on to anyone you let use the model. Later DeepSeek models (V3.1 and up, V4, R1) are MIT. DeepSeek License |
| Qwen2.5 (Alibaba) | Qwen License Agreement | A product with more than 100 million monthly users needs a licence from Alibaba. If you train or improve a model with it, show “Built with Qwen” or “Improved using Qwen”. Qwen License |
| Qwen3.8-Max (Alibaba) | Qwen3.8-Max License | Very large products (over 100 million monthly users or 20 million US dollars a month) must show the model name. Most Qwen3 models are Apache 2.0. Qwen3.8-Max License |
| Qwen3.8 Flash (Alibaba) | Hosted model by Alibaba | Qwen3.8 Flash: hosted model by Alibaba; use is subject to the Alibaba Cloud acceptable use policy. |
| Kimi K3 and Kimi K2.6 (Moonshot AI) | Kimi K3 License; Modified MIT | Very large products (over 100 million monthly users or 20 million US dollars a month) must show the model name in the product. Kimi K3 License · Kimi K2.6 licence |
| GLM-5.3 (Z.ai) | GLM-5.3 License | Keep the copyright notice. Other GLM models are MIT. GLM-5.3 License |
| Step-5 (StepFun) | StepFun Model License 1.0 | Very large products (over 100 million users or 20 million US dollars a month) must show “Powered by StepFun AI”. StepFun Model License |
| NVIDIA Nemotron (NVIDIA) | NVIDIA Open Model License, NVIDIA Nemotron Open Model License, NVIDIA Open Model Agreement, or OpenMDW 1.1, as listed on each card | Use must be consistent with NVIDIA's Trustworthy AI terms. Do not remove or weaken a model's safety guardrails without an equivalent one. NVIDIA Open Model License · NVIDIA Trustworthy AI terms |
| Apache 2.0 and MIT models (for example gpt-oss, Gemma 4, most Qwen3, GLM, Mistral, MiMo, Ling, Beam, Muse Glimmer) | Apache License 2.0 or MIT | No use policy is passed on. Use model names only to describe where a model comes from. Apache License 2.0 · MIT licence |
Notices
- Built with Llama.
- Llama 3.1 is licensed under the Llama 3.1 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.
- Llama 3.3 is licensed under the Llama 3.3 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.
- Llama 4 is licensed under the Llama 4 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.
- Built with MiniMax M3.
- Built with MiniMax M3.1-Flash.
- Gemma is provided under and subject to the Gemma Terms of Use found at ai.google.dev/gemma/terms.
- Qwen is licensed under the Qwen LICENSE AGREEMENT, Copyright (c) Alibaba Cloud. All Rights Reserved.
- Licensed by NVIDIA Corporation under the NVIDIA Open Model License (and, for the models that carry them, the NVIDIA Nemotron Open Model License and the NVIDIA Open Model Agreement).
- Model names are trademarks of their owners. Tīrtha is not affiliated with or endorsed by any model maker.