Open-source AI and Llama hub
Everything to pick, install and run open models on your own computer.
Open modelsWill it run?Setup guidesLicensesHardwareCheat sheetsProjectsCompare modelsGlossaryRoadmapInterview Q&AFree resources
Open models
An open model is one whose weights you can download and run yourself, on a laptop, a desktop or a rented GPU, without sending your text to a hosted service. This table lists each model with its size, license and the memory it needs. Read the license column before you build anything commercial, because "open" does not always mean "free to use for business".
| Model | Maker | Size | License | Context | Best for | Commercial | Memory (4-bit) |
|---|---|---|---|---|---|---|---|
| Gemma 2 9B | 9B | Gemma Terms of Use | 8K | Compact general chat for machines with 8 to 16 GB of memory | Allowed under the Gemma terms; read them | 6.9 GB | |
| Llama 3.1 70B | Meta | 70B | Llama 3.1 Community License | 128K | Workstations or servers with a lot of memory | Allowed under the Meta license terms; read them | 43.5 GB |
| Llama 3.1 8B | Meta | 8B | Llama 3.1 Community License | 128K | General chat on a laptop with 8 GB or more of memory | Allowed under the Meta license terms; read them | 6.3 GB |
| Mistral 7B Instruct v0.3 | Mistral AI | 7B | Apache 2.0 | 32K | Small local assistants, with a permissive license | Yes | 5.7 GB |
| Qwen2.5 7B Instruct | Alibaba Cloud | 7.6B | Apache 2.0 | 32K (128K with YaRN) | Multilingual chat and coding on a laptop | Yes (Apache 2.0 for this size) | 6.1 GB |

