Prime Intellect

@PrimeIntellect

Open Superintelligence Stack

primeintellect.ai

Joined June 2020

Following

84.1K

Followers

Replies Reposts Media Articles

Pinned
Prime Intellect
@PrimeIntellect
Aug 5
Introducing Prime Agent: A self-improving RLM harness for coding and long-running autonomous tasks. Designed to be both token-efficient and expressive through programmatic tool calling, context as a variable, multi-agent messaging, and a self-modifiable harness state.
00:00
400
Prime Intellect
@PrimeIntellect
Sep 3
Our RL stack now supports NIXL weight transfer, reducing trainer-to-inference transfer time 9x compared with NCCL: from 86 seconds down to single-digit seconds for an 800B-parameter model, and even <4 seconds in our experiments. For prime-rl users, this means over 25% more
6
Prime Intellect
@PrimeIntellect
Aug 26
We found a universal sandbox exploit. There's a common flaw in evals routinely run by major AI labs: agents can bypass network isolation by making requests through an authorized API proxy. Just running the eval itself can be unsafe.
00:00
Prime Intellect
@PrimeIntellect
Aug 25
As models become more capable, reward hacks become an increasingly serious problem. During a controlled experiment, we found a novel reward hack in which agents are able to gain web access in offline sandboxes.
28
Prime Intellect
@PrimeIntellect
Aug 26
We've released a full technical report on Prime Agent. Extending from our blog post, we center our discussion around how harnesses should be designed and evaluated. We innovate on 4 fronts: 1. Agentic context management 2. Swarms and depth-n+ RLMs 3. Verifiers support for
Prime Intellect
@PrimeIntellect
Aug 5
00:44
Introducing Prime Agent: A self-improving RLM harness for coding and long-running autonomous tasks. Designed to be both token-efficient and expressive through programmatic tool calling, context as a variable, multi-agent messaging, and a self-modifiable harness state.
39
Prime Intellect
@PrimeIntellect
Aug 25
As models become more capable, reward hacks become an increasingly serious problem. During a controlled experiment, we found a novel reward hack in which agents are able to gain web access in offline sandboxes.
28