Or is the few extra thousand dollars difference between a base and max-spec MBP still just better spent on literally any other practical option (like different harware, remote hardware, cloud AI subscriptions or credits). Or wait to see if there will be an M5 Studio or what inferencing performance next year's M6 may bring?

I am tempted, but even with some new models getting skinnier and more efficient, I am not sure moore's law and the M5 generation is quite there yet to be worth the trouble?

What call's are ya'll making and why?

Comments URL: https://news.ycombinator.com/item?id=47566687

Points: 11

# Comments: 12

Problems I'm seeing: - spending 80% of time troubleshooting the agent framework, vs doing constructive projects - agents writing without reading first - concurrency issues, agents reading from stale memory - agents not consistently backing up before chantes - no built-in safety controls for rolling back to last good config - intallers or plugins stomping all over the file - schema constantly changing and agents not checking docs and then hallucinating syntax - basic problems with json (no comments, no types, very fragile to syntax errors) - even small errors will block gateway from restarting and block the agents from being able to run to autonomously troubleshoot or fix their own mistakes. - openclaw doctor seems to cause as many problems as it fixes

I find myself setting up a lot of scripts, scaffolding, and external AI tools, just to babysit an agentic framework. Which sort of defeats the purpose of using a "ready-made" framework in the first place? Or is there an easier way?

Comments URL: https://news.ycombinator.com/item?id=47565763

Points: 3

# Comments: 0

Maybe there's a very latency sensitive need to send realtime targeting information to a tomahawk missile in flight? But it's also too bandwidth, compute or cost intensive to send a firehose of raw spy-satellite data to a disposable one-way attack munition?

The data centers in space are actually for the spy satellites to use. That's all I got for practical applications.

In all seriousness, I am quite concerned for what the base model and upgrade pricing is going to be for the full M5 line, given the recent AI-driven DRAM supply apocalypse.

Meanwhile, a built-up and elevated stern 'castle' is advantageous place to put the steering and command position, close to the rudder and with visibility of the whole ship, it's rig, plus where the ship is going. While maximizing mid-ship area for cargo. If you have to pick one end or the other, stern is the more comfortable end of the ship being most sheltered from wave action and weather. Being elevated and fortified also helps as a fighting/defensive position, but that is less important for modern cargo ships. 'Anticipation' isn't quite the right word as shipbuilders have always worked within the same basic design considerations and trade-offs, as the sea itself continues to enforce the same fundamental constraints.

Meanwhile in Minneapolis, the overwhelming majority of violent crimes (including aggravated assaults, theft, murders and sexual assaults) are being committed by ICE agents.

So to say that Zen 6/7 supports AM5 on desktop, doesn't necessarily exclude that Zen 6/7 product family in general doesn't support other new/interesting sockets on desktop (or mobile) also. Maybe products for AM6 and AM5 from the same zen family.

Medusa Halo and the Zen7 based 'Grimlock Halo' version might be the interesting ones to watch (if you like efficient Apple-stlyle big APUs with all the memory bandwidth)

But how much of each product they release also just a factor of how much they are willing to spend on inference per query in order to stay competitive?

I always wonder how much is technical change vs turning a knob up and down on hardware and power consumption.

GTP5.0 for example seemed like a lot of changes more for OpenAI's internal benefit (terser responses, dynamic 'auto' mode to scale down thinking when not required etc.)

Wondering if GPT5.2 is also case of them in 'code red mode' just turning what they already have up to 11 as a fastest way to respond to fiercer competion.

If you think of openAI like a new google, as in a new category-defining primary channel for consumers to search and discover products. Well, 2% does seem pretty low.