---
title: "Model licences in this catalogue"
canonical: "https://router.xark.io/licences"
description: "What each licence in the catalogue permits, what it requires, and which models carry it — across 7 families."
section: "core"
updated: "2026-09-02"
source: "https://router.xark.io/licences.md"
---

# Model licences in this catalogue

Apache 2.0 and MIT account for 18 of the 24 models in this catalogue, and impose nothing on a commercial user beyond keeping the notices. The other 6 are spread across 5 licences whose terms are set by their own text, and those are the ones a launch review has to read.

**Availability:** 7 of the 24 catalogued models are callable today. The rest are listed at their intended rate but no configured upstream serves them, so a request naming one returns `no available channel`. Every table below marks which is which.

## Families

| Licence | Models | Callable | Commercial use | Canonical text | Document |
| --- | --- | --- | --- | --- | --- |
| Apache 2.0 | 11 | 2 | Yes | [text](https://www.apache.org/licenses/LICENSE-2.0) | [read](https://router.xark.io/licences/apache-2-0.md) |
| MIT | 7 | 3 | Yes | [text](https://opensource.org/licenses/MIT) | [read](https://router.xark.io/licences/mit.md) |
| Modified MIT | 2 | 2 | Yes, with conditions | — | [read](https://router.xark.io/licences/modified-mit.md) |
| Llama 4 Community | 1 | 0 | Read the text | — | [read](https://router.xark.io/licences/llama-4-community.md) |
| LTXV Open Weights | 1 | 0 | Read the text | — | [read](https://router.xark.io/licences/ltxv-open-weights.md) |
| Stability Community | 1 | 0 | Read the text | — | [read](https://router.xark.io/licences/stability-community.md) |
| Tencent Hunyuan Community | 1 | 0 | Read the text | — | [read](https://router.xark.io/licences/tencent-hunyuan-community.md) |

## What each verdict means

- **Yes** — A standard permissive text that allows commercial use outright, with no user threshold and no field-of-use restriction.
- **Yes, with conditions** — The standard text this licence is built on allows commercial use. The publisher's modifications are not summarised here and may add conditions.
- **Read the text** — A publisher's own licence. It permits third-party hosting, which is why the model is catalogued at all; every other term is in the document.

## Per model

| Model | Licence | Commercial use | Callable |
| --- | --- | --- | --- |
| Kimi K2.6 | Modified MIT | Yes, with conditions | Yes |
| Kimi K3 | Modified MIT | Yes, with conditions | Yes |
| GLM-5.2 | MIT | Yes | Yes |
| GLM-5.2 Air | MIT | Yes | No |
| DeepSeek V4 Pro | MIT | Yes | Yes |
| DeepSeek V4 Flash | MIT | Yes | Yes |
| Qwen3 Max Instruct | Apache 2.0 | Yes | No |
| Qwen3 235B A22B | Apache 2.0 | Yes | No |
| MiniMax M2 | MIT | Yes | No |
| Llama 4 Maverick | Llama 4 Community | Read the text | No |
| Mistral Large 3 | Apache 2.0 | Yes | No |
| GPT-OSS 120B | Apache 2.0 | Yes | No |
| Wan 2.2 T2V A14B | Apache 2.0 | Yes | No |
| Wan 2.2 I2V A14B | Apache 2.0 | Yes | No |
| LTX-2.5 | LTXV Open Weights | Read the text | No |
| HunyuanVideo 1.5 | Tencent Hunyuan Community | Read the text | No |
| CogVideoX-5B | Apache 2.0 | Yes | No |
| Mochi 1 | Apache 2.0 | Yes | No |
| FLUX.2 [schnell] | Apache 2.0 | Yes | No |
| Qwen-Image | Apache 2.0 | Yes | Yes |
| Stable Diffusion 3.5 Large | Stability Community | Read the text | No |
| HiDream-I1 | MIT | Yes | No |
| Qwen3 Embedding 8B | Apache 2.0 | Yes | Yes |
| BGE-M3 | MIT | Yes | No |

## Models we do not resell

| Model | Publisher | Why not |
| --- | --- | --- |
| Sora 2 | OpenAI | Closed weights — no resale license exists |
| Veo 3.1 | Google DeepMind | Closed weights — no resale license exists |
| Runway Gen-4 | Runway | Closed weights — no resale license exists |
| Kling 2.5 | Kuaishou | Closed weights — no resale license exists |
| Wan 2.5 / 2.6 | Alibaba | API-only — weights never released, despite the open Wan 2.1/2.2 lineage |

This is a reference, not legal advice. It exists so you can arrive at your own counsel with a sharper question than whether you are allowed to use the model.