Open Issues Need Help
View All on GitHub Help wanted: test on a real 64 GB Mac 11 days ago
help wanted
MeldlabsAI/meld-turbo MeldlabsAI/meld-turbo A 125B-parameter model, fully local, on a MacBook: Qwen3.8-Flash-Next at ~45 tok/s on Apple silicon.
Open project →
1
A 125B-parameter model, fully local, on a MacBook: Qwen3.8-Flash-Next at ~45 tok/s on Apple silicon.
Python
#apple-silicon#inference#llama-cpp#llm#local-llm#macos#metal#mixture-of-experts#qwen#speculative-decoding