@iuditg Because no open source model is anywhere close in real life scenarios to frontier, and there isn't yet a good enough way to route tasks based on what these cheaper models are good at. Most of the time it's not acceptable to go down 20-30% in success rates
ehhh 25 mil isnt all that much. if I combine my home and work accounts, that's a regular day. if stars align and/or i'm starting a new project without a big blocking graph, I can run like 50-100 mil.
I usually have workers limited to 3 though so it's easier to control, and usually it's me that's a bottleneck
I dunno maybe, I only used to clean up PRs cuz there's something busted on the code review billing on my account where it'll stall saying it ran out of credits, but the account shows no usage. I just basically told it to finish review loops (i.e. review/fix until no P1/P2) on all outstanding PRs, which there were like 10
Ehh goals are broken, I tried yesterday, it spun up 4 subagents which all got errored due to missing credentials immidiately. Didnt do anything for 3+ hours until I asked it for status, it then decided to fix in main thread (took 4 or 5 min) and reported almost 4 hour goal execution
it's most likely that bluetooth was secondary, as in something was happening and they need to fail back to either some sensor/system where bluetooth would interfere or there was a security warning (like possible individual with a bomb or some malicious device in their checked in luggage that is triggered by bluetooth. And since they were unable to verify they basically went back to ground, probably the safest thing to do
@FollowIncentiv If you worked in Burleson or nearby sure, but if you're going to business centers, like say downtown Dallas that's a 1-1.5 hour commute each way. But then again Celina and Prosper is too and they get ridiculous prices. This house's quality is likely 5x better than the new builds
Because those plans don't exist for business in the most part. So Claude does have a Teams Premium at 150, but you blow through those liits pretty fast in a workday, so you cut into your extra usage. Openai only has a $30 dollar team plan, wanna know ho fast those limits go? Pro plans are only individual, and for a company to allow a dev to bring their own software can be disastarous
This is not even counting things like code review tools, auto builders, etc which burn tokens faster than a nuclear plant lol
Fully right but really anyone in management had the same view. Problem has always come down to managing priorities, my problem is I don't have money today, the junior crisis wont materialize until 4-5 years, gives me 4-5 years to figure out how to deal with it. Far from the the right way of thinking, but we do need to consider reality
@AlexRoseJo@aditiitwt I think he means using the hot water to produice steam to drive turbines which will regenerate the energy to drive the gpu. kinda like regenrative braking. I'm assuming thermal physics wont work
@brookejlacey You arenโt wrong that if this is somehow true it will set a nasty precedent. But I think it will self select out. I mean after all funding is a start not the end. Customers will see right through it over time
@thsottiaux yes, so long as it's cheaper. All my stuff is on my own version of symphony, so if you give me a queue I can just send it and yall process whenever you're good and ready
@Gust875250@Ximena_Noir Really? Are you going forward or reversing? I know my f150 had the reverse braking, that one was insanely hard braking. But f150, tacoma (on auto cruise) and Kia all had around 30 mph warning before kicking in