Loading activity…
@svpino
Computer scientist. YouTube: https://youtube.com/@underfitted
AI
philhchen says Fable 5.2 lifts Claude Tag above agent rivals, while the same model’s API setup still scores higher.
AI
Expo’s paid-plan service gives agents cloud iOS and Android test environments, with recordings and browser-based handoffs, the company says.
AI
According to @omarsar0, Keenable built its own search stack and uses agent queries to improve it.
AI
Posts describe agents sharing files and logins on one cloud computer, though access and pricing remain unverified.
AI
Omarsar0 reports zero data retention for Concentrate AI’s BAA-covered routes, but advises checking model and provider coverage.
AI
According to @kimmonismus, Thorsten-4’s predictions were checked against real behavior after four years of repeated interviews.
AI
Posts pitch Nod as n8n Labs’ experimental agent for browser chores, with provider choice and Windows 11 preview support.
AI
IFM’s PAN 2 aims to preview actions as video, but reportedly completed only 34.0% of plans in IFM’s tests.
AI
Steve is reportedly designed to record what went wrong and simulate the outcome of a decision before acting.
Technology
PCMag praises the 16-inch MacBook Pro’s tandem OLED display, thinner build and battery life.
AI
The benchmark tests agents' research methods and software use; every model reportedly scored zero on its end-to-end research track.
AI
TwIL-8B's open-source release drew praise from a commentator who sees locally run models aiding scientific discovery.
AI
Aether's extra credits cost $0.10 for 12 minutes on a small machine, with idle time free, a user says.
AI
LiberAI's model reportedly completed 36% of 600 attempts across 60 new tasks without demonstrations of those tasks.
AI
The managed inference product is pitched for spiky search, OCR and agentic workloads, with a $500 launch-credit offer.
AI
The company also introduced two research previews and described a new Sequoia-led funding round.
AI
Sage2's extraction feature returns structured data from text, while decision answers update as new text arrives.
AI
It returns probabilities for possible answers to audio questions; Hume is inviting access requests through a waitlist.
AI
The Turing TTS team says even the best model it tested fooled listeners only about 10% of the time.
AI
Skydive describes assigning agent teams tasks through Slack, email, iMessage or the web.