Open Superintelligence Stack
- Tweets 3,049
- Following 46
- Followers 84,798
- Likes 44,000
Pinned Tweet
Introducing Prime Sandboxes:
MicroVM sandboxes purpose-built for RL training.
Model training requires running tens of thousands of concurrent sandboxes, leading to complex and costly configuration. We built Prime Sandboxes for our own team. Today we're releasing them publicly.
41
76
721
106,632
Prime Intellect retweeted
every individual
every company
every nation
needs to own their intelligence
Prime Intellect
12
50
356
15,331
Prime Intellect retweeted
Progressing towards open superintelligence 🫡
> multi-agent RL and systems: primeintellect.ai/blog/multi…
> autonomous AI research, from data, pre to post-training primeintellect.ai/blog/measu… & primeintellect.ai/auto-nanog…
> ricursive self improving agent harness: primeintellect.ai/blog/prime…
> continual learning (more soon)
Superintelligence is gonna emerge out of trillions of self improving agents instead of one god model
4
15
159
14,212
Prime Intellect retweeted
Superintelligence is gonna emerge out of trillions of self improving agents instead of one god model
72
56
751
84,702
Prime Intellect retweeted
lots of categories emerging around the AGI stack:
- open model RLaaS
- data & evals
- inference
- coding harnesses
- GPUs
- agent tracing
- sandboxes
at @primeintellect, we agree. we do all of these things. people used to ask why we do so many things. they ask that less now.
12
28
573
24,107
Prime Intellect retweeted
Highest throughput GLM 5.3 🫡
docs.primeintellect.ai/infer…
14
15
221
22,843
Prime Intellect retweeted
We’re hiring a full-time engineer to help build Prime Agent :)
Own features end to end across agent harnesses, cloud agents, and self-improving loops -- integrated with our sandboxes, evals, and hosted training stack.
Full-time, in person in SF: primeintellect.ai/careers/85…
27
12
294
15,760
Prime Intellect retweeted
prime agent v0.9.6 is out:
◆ Support for GPT-6 Sol, Opus 5.5, and Grok 4.7
◆ /mcp plugin catalog with one-click connections to Linear, Notion, Posthog, Stripe, and 60+ more services
◆ Huge perf and reliability pass 🫡
Lots more coming soon :)
24
13
192
12,285
Proud to deliver the most efficient microVM sandboxes on the market at the most competitive prices for our users.
Made by RL teams, for RL teams. Deploy Prime Sandboxes to perform a training run, generate synthetic data, and run an eval or a persistent/remote agent.
Huge push by @a_kirillo and @damian_b 🔥
Introducing Prime Sandboxes:
MicroVM sandboxes purpose-built for RL training.
Model training requires running tens of thousands of concurrent sandboxes, leading to complex and costly configuration. We built Prime Sandboxes for our own team. Today we're releasing them publicly.
2
3
81
7,640
Prime Intellect retweeted
Today, we're releasing Prime VM Sandboxes, co-designed with our research team for large-scale agentic RL training with tens of thousands concurrent sandboxes.
Introducing Prime Sandboxes:
MicroVM sandboxes purpose-built for RL training.
Model training requires running tens of thousands of concurrent sandboxes, leading to complex and costly configuration. We built Prime Sandboxes for our own team. Today we're releasing them publicly.
6
11
192
10,165
Prime Intellect retweeted
Excited to launch Prime Sandboxes: microVMs for large-scale agentic RL training.
We couldn’t find sandboxes that could handle the scale of our RL training, so we built our own.
nitter.net/PrimeIntellect/status/…
Prime Intellect
Introducing Prime Sandboxes:
MicroVM sandboxes purpose-built for RL training.
Model training requires running tens of thousands of concurrent sandboxes, leading to complex and costly configuration. We built Prime Sandboxes for our own team. Today we're releasing them publicly.
11
9
138
7,528
Introducing Prime Sandboxes:
MicroVM sandboxes purpose-built for RL training.
Model training requires running tens of thousands of concurrent sandboxes, leading to complex and costly configuration. We built Prime Sandboxes for our own team. Today we're releasing them publicly.
41
76
721
106,632
Read our full blog post here primeintellect.ai/blog/sandb…
1
31
2,091
Visit our sandbox page primeintellect.ai/sandboxes
29
1,687
The Goodfire team used Prime Intellect to train activation probes to detect reward hacking.
With them, they are able to catch reward hacking in various models.
Their probes are performing similarly or better than frontier LLM-as-judge setups, while being more efficient.
Models know when they’re reward hacking. But they still do it a ton - in 50-96% of rollouts we studied!
We built activation monitors that detect the behavior behind the Hugging Face hack in real time. This can help us stop hacks now - and train future models that don’t cheat. 🧵
11
41
319
34,855
Read the case study here: primeintellect.ai/case-study…
2
20
2,472