Llama 4 Maverick
Llama 4 Maverick
The open-weights default: good enough, cheap to serve, supported everywhere.
Updated
The read
Maverick is the mixture-of-experts middle of Llama 4 and the model most self-hosting teams land on. It is multimodal, it activates a small fraction of its parameters per token so it serves affordably, and every inference server, quantisation format and fine-tuning toolkit supports it on day one. That ecosystem, rather than the benchmarks, is the reason to pick it.
Where it shines
- Supported by every serving stack and quantisation tool that matters
- Mixture-of-experts design keeps serving costs down
- Largest fine-tuning community of any open family
Worth knowing first
- Chinese open-weights rivals now beat it on most benchmarks
- The community licence carries a user-count threshold
Index score
80.4
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
- 23 of 53
- Builds
- LLMs
- Category
- Open weights
- Output
- Downloadable weights under Meta's community licence
- Pricing
- Free tier
- Best for
- Self-hosting a capable multimodal model affordably
Scores are our own editorial judgement, weighted craft 55, speed 10, control 15 and value 20.
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
- 12Qwen 3.6Qwen 3.6ProvisionalThe open Qwen release most people should actually run, in sizes that fit real hardware.Open weightsFree tier83.3
