|
|
|
|
|
|
by wruza
506 days ago
|
|
|
Just get a graphics card and run a prompt-compatible llm yourself. Recent models like phi-4 show decent results (relative to your general amazement baseline) even on medium quantization. I’m running q4_k_m (8gb) with custom “just print and stfu” characters and rarely reach Claude anymore. |
|
|