Tomasz Wojewoda

I spent $6,000 on a Mac Studio to run AI models locally. Then I did the math on $DIEM. Same money. Same model. More tokens. No hardware. A thread 🧵 https://t.co/UYIIybijio

Mar 18

But for raw output per dollar? $DIEM: 6.32M tokens/day Mac Studio: 4.75M tokens/day Same $5,999. Same model. DIEM wins by 33%. And it scales to every model on Venice without buying new hardware. 🫡

Mar 18

The setup: Mac Studio M3 Ultra, 256GB RAM — $5,999 Qwen 3.5 35B running locally via MLX Speed: ~55 tokens/sec output vs. 7.9 $DIEM tokens at $760 each — $5,999 Same Qwen 3.5 35B on Venice API $7.90/day of inference credit. Forever.

Mar 18

@AskVenice @Greensmokegroup @solanamobile Fair price for DIEM is about $2500 to $5000. Perhaps more if token prices keep coming down.

Mar 21

@ErikVoorhees Using it every day. Sabrina in Discord got day 1 issues solved in less than 24 hours. Been solid since then. Venice is the place to use GLM 5.2 privately for $20 per month. Unless you have $500K to run it on a dedicated DGX.

Jun 23

@AskVenice @Zai_org Got to give it a shot!

Jun 16

Cooking on @AskVenice https://t.co/9DQHKRczXG

Jun 18

@AskVenice @ErikVoorhees @Kimi_Moonshot Testing

Jun 16

venice.ai is also an option.

Feb 28

@AskVenice why are you capping GLM 5-2 on max_completions? Let it free please.

Jun 18