We recommend using Prime Agent for both day-to-day work and standalone evals. We find strong downstream performance gains for open and closed frontier models using Prime Agent on standard long-context and long-horizon tasks, despite modern models not being trained around Prime Agent.

MIT PhD student Alex Zhang on why simple harness abstractions beat models trained around their own harness: "When these abstractions are implemented in a really simple way, basically what you see with Prime Agent is that you can slot in any model. We don't train any model around Prime Agent." "In a lot of cases, we see Prime Agent doing better than models trained around a certain harness. It's purely because these abstractions take advantage of the raw capabilities of the models." @a1zhang @sethkarten @PrimeIntellect

Prime Agent is a general-purpose coding harness On ARC-AGI-3, it scores 95.5%, surpassing the human-expert baseline, but the gain is not benchmark-specific. We see major improvements across models when compared to their proprietary harnesses:

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.

fun fact: during our team's trip to pytorch conf in paris we had a heated debate on whether we should use bash or python as the main repl for a new training harness we were designing. at some point it occurred to us that we could have the best of both worlds with ipython there is something oddly satisfying in an agent interacting with the whole world through the universal language that is code. i am sure your agent will appreciate it too!

ARC-AGI 3 is now completely saturated (solved) by a new AI harness called Prime Agent! It achieved its highest score with Opus 5, but it can work with any other model too. It's especially well suited for agentic coding tasks. Incredible achievement by @PrimeIntellect 🔥

very excited about this. not all frontier models fully use the capabilities yet but seeing the model create its own functions/"language" and refine itself is pretty cool. here is kimi K3 building abstractions to launch new runs on the nanogpt optimizer track

SITUATION DETECTED: Prime Intellect is releasing Prime Agent, a self-improving harness for coding and long-running autonomous tasks. The team reports 95.5% on ARC-AGI-3, above the human baseline, and says the gain is not benchmark-specific.

Excited to release Prime-Agent A self-improving RLM harness for coding and long-running autonomous tasks. Our goal is to enable everyone to build and use self-improving agents; and the harnesses, models, and infrastructure to support them. Prime-Agent is token-efficient and expressive through programmatic tool calling, context as a variable, multi-agent messaging, and a self-improving harness state. It combines three ideas we’ve been exploring: >Recursive Language Model–native programmatic tool calling >Persistent multi-agent orchestration >A self-improving continual harness Together, they let the model act on, and improve its own context and harness. @kevinjosethomas @a1zhang @sethkarten and team cooked 🙌 Install it: curl -fsSL | sh

The boys at prime intellect have cooked. Their Prime Agent is a general purpose coding harness And on ARC-AGI-3, it scores 95.5%!! “surpassing the human-expert baseline, but the gain is not benchmark-specific.” Prime intellect is hyper RL & CL pilled.

New AI agent Scores 95.5% on ARC-AGI-3 The pace is becoming absurd ( and I love it) "general-purpose coding harness" Prime Agent combines three ideas: "1. Recursive Language Models-native programmatic tool calling 2. Persistent multi-agent orchestration 3. A self-improving Continual Harness"

Super exciting: Prime Intellect launched Prime Agent, an open-source coding harness that turns long-running AI sessions into a programming problem. Its only tool is a persistent IPython kernel. The model can programmatically search its history, call tools, launch persistent sub-agents and store useful state outside the active context. Prime Intellect: "The RLM treats context as a variable and subagent delegation as function calls inside a REPL." They reportstrong gains across long-context and long-horizon tasks. With Opus 5, Prime Agent scored 95.5% on ARC-AGI-3, narrowly above the benchmark’s reported 95.4% human expert baseline. It also built working SEGA Genesis and Game Boy Color emulators from scratch in Rust on a preview benchmark :D

we are releasing our own coding agent, I have been using it daily since a month now and my favorite place to work with open and close model 1. RLM: * programmatic tool calling * persistent ipython kernel across compaction for long term memory * Subagent daemon and UI 2. Self improving continual harness, model have access to tool to programmatically edit their own harness and skills

Prime Agent combines three ideas: 1. Recursive Language Models-native programmatic tool calling 2. Persistent multi-agent orchestration 3. A self-improving Continual Harness Together, they let the model act on its own context and harness

Prime Agent combines several research ideas we've been exploring over the past few months: - Recursive Language Models-native programmatic tool calling - Persistent multi-agent orchestration ⁃ A self-improving Continual Harness Try it out: