@tamarajtran@theo how has no one mentioned yet that this is just Frankenstein version of @morphllm Compact? At least that model has a bigger context window and was explicitly trained for this.
How is he wrong?
(In this instance he is of course wrong)
The point is abysmal security culture. If we get a frontier model “for free”, how many will run it with —dangerously-skip-permissions overnight? Hell, I’m doing this with Ox.
@drbarnard@CoachDanGo Any profit was coincidental. They insisted on paying me so I threw together a silly little pricing calculator. I probably profited like $40 for nearly 40 hours of work.
@montecristocap I’m so lost in the sauce! It took me a sec to realize “(GPT6 / Ant equivalent)” was referring to Anthropic not Ant Research aka inclusionAI. (makers of the Ling and Ring models)
I dunno, anytime a new Chinese model drops, X goes into hysteria, then goes right back to using OpenAI/Anthropic models. Wake me up when I should actually care.
I felt a great disturbance in the Force, as if millions of voices suddenly cried out in terror and were suddenly silenced.
I fear something terrible has happened.