Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
It’s the first Human Interaction Model (HIM).
Community note
The 48% figure and "video Turing test" claim are from Tavus's own study of 54 one-minute calls, not independently verified or using a standard protocol. Griffin-Lite leads NVIDIA's VideoFDB benchmark on their public leaderboard.
cellcog.ai/blog/tavus-gri…
research.nvidia.com/labs/amri/proj…
tech-ish.com/2026/10/02/tav…
3,231
4,246
38,806
19,480,350
COMPUTE THINGS
Excited to announce Volantis's $88M Series A.
We are solving Al's memory bottleneck by using optics, enabling chips with huge amounts of fast & cheap memory.
By boosting both the memory bandwidth and capacity per chip by orders of magnitude, we enable ultra-fast inference (up to 10,000 tps/user) for large models (>10T) - with low $/tok to boot.
Initially, this will enable insanely fast agents - think coding agents that finish in minutes or even seconds instead of hours.
More excitingly, optics is a fundamentally scalable way to increase memory systems. Not 2X/year, but by orders of magnitude across new generations. This will enable a structurally new Al industry, including restarting scaling laws, holding entire repos in context windows & more.
Our team has pioneered many core semiconductor technologies: the 1st CoWoS product, early HBM, the 1st silicon photonics CPO systems, the 1st high volume tunable VCSELs, the 1st processors to directly communicate using light & more.
We’ve already sent data >10× farther than equally tiny electrical wires inside a chip package. Our next iteration is already taped out and targets world-record bandwidth density over relevant distances, read more: volantissemi.ai/news-insight…
11
431
Solar age inbound
Introducing NetworkOcean.
Our future is bottlenecked by the speed at which we can deploy power.
We’re building floating solar arrays powering data centers at sea. This will soon be the fastest way to deploy new compute, skipping years of land, grid connections, and traditional infrastructure delays.
Our 2030 target is 1 GW/week.
@network_ocean_
2
12
399
up and to the right !
We hit out second $1M revenue month at Lapis, some thoughts on sales / gtm today:
- You can either expand more into one of our existing customers or get more customers. There's no one fixed way to do this but our first 1M revenue month we had a ton of new customers. In our more recent one, we expanded more into our current customers. Both are hard (getting money is always hard) but expansion is ever so slightly easier
- Do not use AI generated sales "scripts". We buy a lot of tools too and it is quite obvious when someone wastes time by following a script. The old script based sales paradigm is over. Learn how to sell on the fly, talk about value.
- The number of buyers for enterprise AI tools is larger than ever before. Competition, funding, etc. literally does not matter. Build a good enough tool and you can sell to buyers just fine.
- A lot of enterprises are having some sort of existential crisis (dependent on team) over how they are going to level up using AI tooling. Instead of a sales motion around replacing things, talk about how you can assist them with your fancy new AI tools and make their work better.
- Speed is everything. Buyer can buy on one call if you do enough prep. What you put in is what you get out. If you're lazily taking calls, customers will lazily follow you around. Be proactive and do a lot of research before the call so the prospect doesn't feel like you're wasting their time
1
3
33
2,249
Solar is already here but I guess you didn’t get the memo
Nuclear is too damn slow to drop our power bills. Americans can’t wait decades.
That’s why states like Texas are choosing solar + batteries as their fastest growing power source. It’s the only power that can be deployed quickly enough at scale.
2
11
1,078
The future of American energy is solar.
It's our grid's fastest growing energy source. In may solar outproduced coal.
But few recognize that it is an AMERICAN invented technology, made feasible by Bell Labs with the silicon cell.
And American solar innovation isn't over
Bullish on solar. It's time to get excited.
@optivolt is embedding power electronics directly inside solar panels. Existing panels are wired in series like Christmas lights. If cells get shaded (by a tree, dirt, bird poop, snow) they choke the current flowing through their section of the panel.
Optivolt's power balancer redistributes energy inside the solar panel at the substring level in real time, letting weaker cells borrow from stronger ones; instead of a shaded cell dragging down its neighbors, the strong cells carry it.
They claim up to 6x more output in shading compared to incumbent installs.
"Sunlight is the most abundant energy source available to civilization yet we somehow capture almost none of what it offers. Solar isn’t intermittent the way people casually assume; it's predictable, cyclical, and, with enough capture and storage, continuous. Difficulties arise only between photons hitting solar cells and electrons reaching the grid."
4
3
22
2,326
nick retweeted
Solar spent a decade as the second-class citizen of American energy: dumb panels on rooftops, made in China, sold on thin margins. The glamorous money went to nuclear instead.
For Arena, @freeshreeda on @optivolt and the case for solar as a technology, not just a commodity.
8
4
42
18,582
nick retweeted
Paid subscribers to @palladiummag who were unable to attend my lecture in San Francisco last week on the lost history of ancient civilization are now able to watch the full video recording online.
Watch here or subscribe to watch:
letter.palladiummag.com/p/ci…
10
15
98
18,107
People underestimate how much stress is accumulated in your body when you repeatedly ignore what you know you need to do
6
32
1,125
VCs will yap at you like a bunch of blinking autists about how their fund thesis is interstellar
2
21
1,496
