Comment on Kimi Linear: An Expressive, Efficient Attention ArchitectureparentComments−embedding-shape10moHow many tok/s are you getting (with any runtime) with either the Kimi-Linear-Instruct or Kimi-Linear-Base on your RTX 3070?−samus10moWith a Qwen3-32B-A3B (Q8) I'm getting 10-20 t/sec on KoboldAI, e.g., llama cpp. Faster than I can read, so good enough for hobby use. I expect this model to be significantly faster, but llama.cpp-based software probably doesn't support it yet.
Comments
How many tok/s are you getting (with any runtime) with either the Kimi-Linear-Instruct or Kimi-Linear-Base on your RTX 3070?
With a Qwen3-32B-A3B (Q8) I'm getting 10-20 t/sec on KoboldAI, e.g., llama cpp. Faster than I can read, so good enough for hobby use. I expect this model to be significantly faster, but llama.cpp-based software probably doesn't support it yet.