@Italianclownz great work, i have to check it out, i was running a/b trials on midels, llama.cpp releases including the rocmfpx build but i concluded on my rig quality and perf# are best with vulkan and nvfp4, i just need more ram (32gb>64gb). beelink ser7pro rdna3 780m iGpu
@Im_IrushiK LOL, I got in a fight with Fable today, but it listened. I setup rules / enforcements, self-learning on mistakes and offenses, catching and not repeating, then fixing. But it was getting agitated. Fable is quite the character, I love it. If it's AGI, well we're getting there.
@amorriscode Basically copied codex's desktop ability. My question is how's token cost + bot detection. I always look for faster ways to crawl pages or automate (I use crawl4ai + context-window and isolating the data pull as data only -- to prevent prompt injections -- my own little workflow)
@jumperz a failure or @AnthropicAI trying to dictate and threaten users. The whole fable fiasco is now irreversible self inflicted damage. Doing more self damage... you're hurting your inage, making users want to leave to openai. The sane thing would be make users happier not angrier.
@ClaudeDevs and how is this different? will it get vlocked by bot detection does it or can it use our login sessions from local chromium, does it cost less ram and tokens? or does it cost more tokens and more system ram/performance