Building AI agents at Meta. Opinions are my own.
- Tweets 1,209
- Following 162
- Followers 76
- Likes 4,841
P a c i n g
Rumors I've been hearing: Bel from OpenAI is an extremely good model, a lot better than Fable 5.5, and it'll probably release by the end of the year, likely December. But that doesn't mean Anthropic is going to lose. I've also heard Anthropic has done a really, really big new pretrain within the last week. This is going to be fun 👀
17
Pachu retweeted
let's do swarms to port a thing to another thing. how much can it cost? $10?
12
2
61
10,336
the swarms are coming
Today we’re launching LiteLLM Lens.
@LiteLLM is already the chokepoint for 100% of your enterprise’s AI traffic. We believe the next era of the gateway is using the data flowing through it to help your agents improve.
We’re building for a future where developers run agentic swarms generating 200K+ traces.
We’re focused on two things ⬇️
h/t @Mokhalil01 @yujonglee
15
Do people not understand that terminal bench is the only one that matters?
Lots of discussion out there about our next model(!), so I wanted to give an early look as soon as possible. Introducing Gemini 4 Argon!
It shows frontier performance in complex workflows, cyber defense and software engineering. Teams are using it extensively at Google, from coding to quantum computing, great feedback.
Here’s a look at the benchmarks:
19
yeah opus 5.5 is clearly dumber than Fable 5.1 in long horizon complex tasks that the benchmarks don't accurately reflect.
14
lesswrong.com/posts/6cb7qd3R…
great write up on swarm scaling by @tobyordoxford . tl;dr scaling up a swarm is a tradeoff of cost for faster results. I think the viability of a swarm will come down to whether the models get fast enough to accomplish long horizon revenue generating tasks on their own. Once we reach that point, investing more to scale the swarm and speed up will be a no brainer. It will also be interesting to see if as models get smarter, they may be able to self coordinate more efficiently, even on tasks humans generally consider hard to parallelize.
16
The swarms are coming
Bot detection & human verification will be one of the most urgent demands for businesses over the coming years. Agent swarms will suffocate every website and form; small companies and government websites are most vulnerable.
There is a huge gap in the market for this right now. When we looked at what offerings were in the market to use at X, there was not a single company that brought together all the latest technologies so we had to do it all in-house.
12
Agent swarms are clearly the future
Fascinating AI swarm dynamics: a few agents spontaneously emerge as highly connected hubs, while most remain locally connected. The swarm develops a strongly heterogeneous interaction topology with a long-tailed degree distribution - an emergent organizational structure arising from initially decentralized local interactions. There is no central planner assigning roles; the swarm builds its own coordination architecture, with information brokers and increasingly global integration emerging from local behavior.
4
55
Pachu retweeted
Periodic sadness reminder that the anthropic subscription still does not permit third party harnesses :(
93
71
1,400
63,965
Do not be scared; rejoice in that you may witness the end times, for in the small blip in the universe that is humanity, it's end will be its only significance.
1
21