Leading research at @arcee_ai | Formerly Data Research Lead @DbrxMosaicAI | Visiting Researcher @meta | Ph.D | #TXSTFOOTBALL fan | linktr.ee/code_star
Redwood City, CA
- Tweets 22,465
- Following 2,020
- Followers 7,722
- Likes 50,515
Pinned Tweet
I’m so incredibly proud of what we are building at @arcee_ai . It’s a truly special place. We are only getting started though.
We’ve been quiet for a while, but we haven’t been idle. I can’t wait to share more about what we have been grinding on for the past ~6 months.
I won’t spoil anything, but you can get some idea from this announcement.
Today, we are announcing our Series B funding round, valuing the company at more than $1B.
This round accelerates our next-gen Trinity models across diverse infrastructure, expands our work with the DOE and national labs on Genesis-Science-1, and enables us to build the platform teams need to build, evaluate, deploy, and operate open models in production.
We are grateful to our team, partners, open-source community, and investors.
Led by @Vista_Equity, Cambium Capital, and @emergencecap, with participation from AI10 Ventures, @Hitachi, IAG, @M12vc, @p7ventures, and @Wipro.
4
7
105
5,096
Cody Blakeney retweeted
we started out the summer trying to build the next generation of our VLM family, world class at classical image and video understanding. as we branched into tool calling capability, datasets which skew heavily towards traditional LLM domains like web search, we discovered some incredible cross task generalization: the model can control physical embodiments like drones and perform real world physical search tasks, demonstrating spatial awareness and navigational skills. this model is incredibly powerful, its fast, and you can throw basically any video, image, or real world agentic task at it. give it a try :-)
Today we're releasing Mk1.5: a new intelligence layer for embodied agents.
It flies drones, controls quadrupeds, powers smart glasses, tracks objects, searches the web, reasons visually, and dispatches its own sub-agents.
One model, no platform-specific retraining. 🧵
5
2
29
15,349
Cody Blakeney retweeted
We're excited to finally release our next generation model for embodied intelligence. We packed a lot into this release, from new architectures (2-4x speed up) to adding audio as a first class modality to new post-training, we're building the foundations for anyone deploying anything in the physical world.
Today we're releasing Mk1.5: a new intelligence layer for embodied agents.
It flies drones, controls quadrupeds, powers smart glasses, tracks objects, searches the web, reasons visually, and dispatches its own sub-agents.
One model, no platform-specific retraining. 🧵
8
11
108
10,785
How many versions of Humanity's Last Exam do you think they are going to release before they need to change the name?
9
61
3,573
Cody Blakeney retweeted
very interesting developments what is this arcee dot ai?
Thrilled to announce I've joined @arcee_ai as Head of Compute / Infrastructure to build the model stack for the next generation of American open source frontier intelligence.
I'm hiring. If you're interested in working on a rigorous team at the frontier of infra, DMs are open!
1
2
27
1,941
Cody Blakeney retweeted
Gerred has had a phenomenal impact on our execution speed in just 4 weeks. Happy he’s here and we’re investing heavily in hires for his team. Consider reaching out.
Thrilled to announce I've joined @arcee_ai as Head of Compute / Infrastructure to build the model stack for the next generation of American open source frontier intelligence.
I'm hiring. If you're interested in working on a rigorous team at the frontier of infra, DMs are open!
1
3
39
4,573
Cody Blakeney retweeted
Keith is the best engineer I ever worked with and we got to hit on all of my favorite topics in this conversation: pure math, chess, engineering, and what all of this means for the kids.
The clankers have the keyboard. Rather than mourn the transition, we find it exhilarating.
Announcing After Coding, a new Pebblebed podcast with @keithmadams. Episode 1 with @josh_wills is live.
link >>
2
11
65
9,737
Cody Blakeney retweeted
Thrilled to announce I've joined @arcee_ai as Head of Compute / Infrastructure to build the model stack for the next generation of American open source frontier intelligence.
I'm hiring. If you're interested in working on a rigorous team at the frontier of infra, DMs are open!
54
21
355
25,059
Cody Blakeney retweeted
The clankers have the keyboard. Rather than mourn the transition, we find it exhilarating.
Announcing After Coding, a new Pebblebed podcast with @keithmadams. Episode 1 with @josh_wills is live.
link >>
1
10
49
16,982
Cody Blakeney retweeted
such unserious extinction hysterics
absolute brain-wormed white knights
pacing theatrics for who, exactly?
rules for thee but not me!
JUST IN: Anthropic is considering releasing a new AI model to counter OpenAI’s GPT-6 Astra momentum, just days after CEO Dario Amodei called for the industry to “slow the pace” of AI development. — Reuters
2
15
263
22,518
Cody Blakeney retweeted
this is why you should be an ultrafinitist
all these kids are getting their brains fried by ideas of negative infinities
Over 2 years ago @sayashk and I wrote a detailed deconstruction of p(doom) and argued that its primary function is to launder vague, evidence-free intuitions and fears through a facade of quantification. It remains 100% relevant today. normaltech.ai/p/ai-existenti…
7
4
110
14,078
Anyone who has had to clean web scrapped code knows your moral obligation to remove the UNBELIEVABLE amount of plain text secrets, passwords, full names, addresses, and dates of birth people just leave everywhere.
The ratio of files that have something like that would horrify you.
think of the children! think of all the keys that will be exfiltrated! they might've been leaked through plaintext before on github, but now they're THERMALLY COORDINATING in morse code!
hide your kids! hide your wife!
and hide your husband cuz...
15
1,193
RT @fujikanaeda: Building NeMo Data Designer was such a crazy ride. Now the team has released the tech report for what we grew it into!
L…
1
40
And I thought Github going down all the time now was bad.
The more popular codex and claude code get, the more people are going to try and target them. OAI and Anthropic have fantastic models and they do things many open models can't, but ... its a lot harder for someone to shut down or intercept your data the models you deploy yourself on your hardware.
If regulation moves us to fewer players, not only does that mean less choice, it means singular points of failure.
The entire way software engineering works at most places has completely changed with coding agents, and if just 1 Company have a bad day it's a going to mess up your road map and timelines.
On July 25, we hacked OpenAI.
Two bugs let us take over ChatGPT/Codex accounts of OpenAI employees (+some unaffiliated users) and reach connected services: Outlook, Slack, GitHub, etc.
We proved it with a PR in OpenAI’s internal codebase . It took us <72h. 🧵
2
14
1,235
My other hot take about this, is the REAL vulnerability here wasn't that agents made it possible to build exploits faster, its that good old social engineering and people (and their choices) were the weak point.
Treating your agent having access to google, slack, and github, while incredibly conveniant is absolutely going to put critical infrastucture at greater risk. The people most vulnerable to these problems are probably the very same people at these labs who are both devleopers of the models, early adopters, experts at using agents, and probably feel like they can take bigger risks than the average person with the level of access and control they give agents because they know how to manage and instruct them so as to prevent the *agent* from doing harm with access to these systems.
Hijacking an agents access to core company systems. Boy there are going to be a lot of security policies that develop around that I'm sure.
88