AI agents have bad taste in long-term decisions and a new benchmark proves it

Technology · en

AI agents have bad taste in long-term decisions and a new benchmark proves it

The Decoder · Oct 18, 2026, 8:49 AM UTC

A new benchmark called Taste-Bench tests how well AI agents make strategic choices during long-running tasks. Even the best model, GPT-5.6 Sol, picks the right path on just under 60 percent of simple binary questions. Thinking longer doesn't help.…

Read original report Visit The Decoder

Technology · Engadget

You don't have to turn off your phone to switch SIM cards (but you probably should)

Swapping SIM cards isn't something you'll do all the time, but when it does come up, you should do it carefully.Read original ↗

Technology · Engadget

iPhone Duo shipping estimates have already been pushed back to 2027

Apple called the early demand "nothing short of extraordinary."Read original ↗

Technology · TechCrunch

Ember’s new temperature-controlled mug makes my nightly tea ritual even better

The biggest change about Ember's new mug is that you no longer have to open the app to adjust your drink’s temperature.Read original ↗

Technology · Engadget

Should you have both Ethernet and Wi-Fi for your home?

Ethernet will keep your devices connecting online in record time, while Wi-Fi keeps them untethered. But how well can they both work together?Read original ↗

Technology · Engadget

How many watts should your Bluetooth speaker have?

If you're confused by Bluetooth speaker wattage, here's what those numbers actually mean. Spoiler: more wattage doesn't necessarily mean a louder speaker.Read original ↗

Technology · WIRED

Here’s Exactly How Tesla Made Its New 230-Mph Car ‘Fly’

It turns out the Roadster 2 both sucks and blows.Read original ↗