Name: I Was Not Expecting This! 120 BILLION Params, 120 Tokens PER SECOND (feat llama.cpp)
Uploaded: 2025-08-08T05:52:52+00:00
Duration: 6 min 39 s
Description: These speeds alone open the door to so many cool things! Not to mention, gpt-oss:20b is a great model, but why not run its bigger brother at comparable speeds! At least some of the time! Dual 5090s gp

1 month ago

Technology gpt-oss gpt-oss:120b gpt-oss:20b llama.cpp llama-server 5090 nvidia cuda AI neovim

These speeds alone open the door to so many cool things! Not to mention, gpt-oss:20b is a great model, but why not run its bigger brother at comparable speeds! At least some of the time!

Dual 5090s
gpt-oss:120b @ 120 Tokens / second
gpt-oss:20b @ 270 Tokens / second

Loading comments...