OMLX 0.7.0.dev4 adds DeepSeek V4.1 CED prefill with up to 79% faster prompt processing on M3 Ultra…
OMLX 0.7.0.dev4 adds DeepSeek V4.1 CED prefill with up to 79% faster prompt processing on M3 Ultra (opt-in), multi-request Lightning MTP, and plenty of quality-of-life improvements including one-click model settings. You can now apply model settings with
Get more done by giving 2 AI agents different jobs.
I run both inside Agent OS: DeepSeek Harness → heavy coding + building Hermes → memory + scheduled background work One dashboard. Each agent does what it’s best at. Want the SOP? DM me.
I know that Astra and Fable are great, but those are just games that can be tolerated by wealthy individuals who hit oil.
With DeepSeek V4.1 Flash, you can play Mario for just $2. 230 million tokens used, 1.78$ for 870 API requests
Wow, I’ve been able to enjoy Astra 6x longer!
By using the oh-my-hermes harness, I managed to save 6x on tokens and get results 4x faster. Oh-My-Hermes: It comes with custom prompting skills and sub-agent routing, so instead of wasting expensive Astra
You have reached the end of the archive
All of deepseek