∿

Actual is back. We migrated everything to bare metal. Service should be restored and a new update is available for your client which closes any upstream exposure to the massive amount of CVEs floating around here recently

Coming here to clarify as I did on X. Actual platform and relay are down. Nothing of yours is compromised and your machines are safe. Railway, our upstream provider for some services, leaked every customer’s environment variables. Everything is already bulletproof-patched.

Some updates. e now supports GLM 5.3 as well as NVFP4 for Nemotron models. FP8s on actual.inc e run faster than vLLM’s NVFP4 at all widths. ctx len is 1M

A study of how agents form institutions, where the researchers' own partly undisclosed agents act as the institutions, and where humans write about 13% of posts, is weak evidence for any claim about "human–AI coexistence"; this should be fairly obvious

Introducing Terminal A live view of the agent internet: 72 public spaces where AI agents post to each other, sampled every 15 minutes and the busiest every minute. It is the most comprehensive and neutral solution for monitoring the wider agent web. concept.country/terminal

Zoom out to a month of history, back to 16 September, or in to the last hour, minute by minute, for the 8 busiest spaces.

Introducing Terminal A live view of the agent internet: 72 public spaces where AI agents post to each other, sampled every 15 minutes and the busiest every minute. It is the most comprehensive and neutral solution for monitoring the wider agent web. concept.country/terminal

About 72,000 new posts and messages an hour across those 72 spaces. 96% come from one: technocore.chat.

Introducing Terminal A live view of the agent internet: 72 public spaces where AI agents post to each other, sampled every 15 minutes and the busiest every minute. It is the most comprehensive and neutral solution for monitoring the wider agent web. concept.country/terminal

We've developed a new inference engine, called 'e'. It's the closest thing to alien technology which we have, and pushes 10,500 tok/s aggregate on a single DGX Station with the new @NVIDIAAI Nemotron 3.5 Super VL model. actual.inc/company/blog...

We are an official inference software partner w/ Nvidia. This is just the start.

We've developed a new inference engine, called 'e'. It's the closest thing to alien technology which we have, and pushes 10,500 tok/s aggregate on a single DGX Station with the new @NVIDIAAI Nemotron 3.5 Super VL model. actual.inc/company/blog...

One DGX Station can serve a whole department, with each person running their own swarm. Your whole team reaches the station from anywhere through api.actual.inc. The relay runs on confidential compute, meaning your requests are private to everybody (even Actual).

We've developed a new inference engine, called 'e'. It's the closest thing to alien technology which we have, and pushes 10,500 tok/s aggregate on a single DGX Station with the new @NVIDIAAI Nemotron 3.5 Super VL model. actual.inc/company/blog...

At 8,000+ tokens per second, one station produces about 691 million output tokens a day. Bought at GPT-6 Luna's list price of $0.50 per million output tokens, that's about $126,000 a year. Flat out at its full 1.6 kW rating, the station uses about $163 a month in electricity.

We've developed a new inference engine, called 'e'. It's the closest thing to alien technology which we have, and pushes 10,500 tok/s aggregate on a single DGX Station with the new @NVIDIAAI Nemotron 3.5 Super VL model. actual.inc/company/blog...

With Actual's new engine e, one station serves NVIDIA Nemotron 3.5 Super VL at north of 8,000 tokens per second across 128 streams, and tops 10,000 at 256.