seagalli

[seagalli @seagalli]

@somehowali Yes, data anomaly classification

[seagalli @seagalli]

@ibocodes Example: tool selection, sentiment, tagging, etc…

Why: sub second response, sub .0001 cost/call, pretty smart

We expect to see improvement in intelligence but it’s decent.

[seagalli @seagalli]

@ibocodes The confusion is that it’s not solving a problem that was not being solved. It’s really good (intelligent, fast, and cheap) at solving that same problem.

The main use case, which we use it for, is agent workflows at scale. That is: context in and get enums out. Fast and cheap.

[seagalli @seagalli]

@Angaisb_ The benchmark for production business value at scale is OpenRouter Rankings. @grok is just fine. https://t.co/56dFGSys6N

[seagalli @seagalli]

@murphsturph TG has the ability to break 20-50 yard run on you with one minor mistake. Once this started to sink in for SDSU defenders the whole defense gets slower and can’t cheat or over play. The problem then compounds once the backfield starts producing and fatigue sets in.

[RUDE @rudeboyfinesse]

ELLIOTT IS REAL. 😌