Skip to main content

πŸ§‘β€πŸ”¬ Models Registry

Completion models (--model)​

We recommend using

  • For 1B to 3B models, it's advisable to have at least NVIDIA T4, 10 Series, or 20 Series GPUs, or Apple Silicon like the M1.
  • For 7B to 13B models, we recommend using NVIDIA V100, A100, 30 Series, or 40 Series GPUs.

We have published benchmarks for these models on https://leaderboard.tabbyml.com for Tabby's users to consider when making trade-offs between quality, licensing, and model size.

Model IDLicense
StarCoder-1BBigCode-OpenRAIL-M
StarCoder-3BBigCode-OpenRAIL-M
StarCoder-7BBigCode-OpenRAIL-M
StarCoder2-3BBigCode-OpenRAIL-M
StarCoder2-7BBigCode-OpenRAIL-M
CodeLlama-7BLlama 2
CodeLlama-13BLlama 2
DeepseekCoder-1.3BDeepseek License
DeepseekCoder-6.7BDeepseek License
CodeGemma-2BGemma License
CodeGemma-7BGemma License
CodeQwen-7BTongyi Qianwen License
Codestral-22BMistral AI Non-Production License
DeepSeek-Coder-V2-LiteDeepseek License

Chat models (--chat-model)​

To ensure optimal response quality, and given that latency requirements are not stringent in this scenario, we recommend using a model with at least 1B parameters.

Model IDLicense
Mistral-7BApache 2.0
CodeGemma-7B-InstructGemma License
Qwen2-1.5B-InstructApache 2.0
CodeQwen-7B-ChatTongyi Qianwen License
Codestral-22BMistral AI Non-Production License
Yi-Coder-9B-ChatApache 2.0

Embedding models​

Model IDLicense
Nomic-Embed-TextApache 2.0
Jina-Embeddings-V2-CodeApache 2.0