Poll: Should mLLMCelltype Support Local Llama Models?

[cafferychen777]

Poll: Should mLLMCelltype Support Local Llama Models?

Background

There have been requests for mLLMCelltype to support local Llama models for cell type annotation. However, many providers (including those accessible through OpenRouter) already offer free access to the latest versions of Llama models.

Current Situation

• OpenRouter and several other providers offer free access to advanced Llama models
• These cloud-based solutions often provide the most up-to-date model versions
• Using cloud APIs eliminates the need for local GPU resources
• Our package already supports multiple providers through OpenRouter integration

Question

Given that cloud providers (including free options) already offer access to the latest Llama models, is there still a need for mLLMCelltype to support local Llama model deployment for single-cell annotation?

Options

• Yes, local Llama support is necessary despite cloud alternatives
• No, cloud-based Llama models through existing providers are sufficient
• Only if it can offer significant advantages over cloud-based solutions

Please vote and share your thoughts in the comments!