Qwen 3.6
Qwen 3.6
The open Qwen release most people should actually run, in sizes that fit real hardware.
Updated
The read
Qwen 3.6 is the current open-weights generation, shipped in a dense mid-size model and a sparse one that activates only a few billion parameters per token. Both punch above their size, both are Apache-licensed, and both are supported by every serving stack worth using. For a self-hosted assistant, a fine-tuning base or an on-premise deployment, this is the most sensible starting point in the index.
Where it shines
- Apache licence with no user or revenue conditions
- Sparse variant runs fast on modest hardware
- Best-supported open family outside Llama
Worth knowing first
- Alibaba now keeps its strongest models closed
- Behind the frontier on the hardest reasoning tasks
Index score
83.3
Provisional. This score is read from the product's public capability, not from the one-prompt rebuild the fully reviewed entries went through, so treat it as a placing rather than a verdict.
- Rank
- 12 of 53
- Builds
- LLMs
- Category
- Open weights
- Output
- Apache-licensed weights, several sizes
- Pricing
- Free tier
- Best for
- Self-hosting and fine-tuning on hardware you already have
Scores are our own editorial judgement, weighted craft 55, speed 10, control 15 and value 20.
Compare it with
Also worth a look
Full index- 05Kimi K3Kimi K3ProvisionalThe largest open-weights model anyone has shipped, and it reads a million tokens without blinking.Open weightsFree tier87.5
- 06GLM-5.2GLM-5.2ProvisionalFrontier-class output, an MIT licence, and a million tokens of context. That combination is the story.Open weightsFree tier87.1
- 14Mistral Large 3Mistral Large 3ProvisionalEurope's answer, published under Apache 2.0, with a quarter of a million tokens of context.Open weightsFree tier82.9
