The Open Superintelligence StackOwn Your Intelligence
Train, deploy, and continuously improve your own models on an integrated compute, training, inference, and sandbox stack.
$ pip install prime
Backed by
Lab. Post-train your own self improving agents
FIG.1
Turn any task into an RL environment. Init, develop, eval, and push with the Prime CLI.
FIG.2
Hosted evaluations for you to benchmark the performance of your models.
FIG.3
Train large-scale models optimized for agentic workflows.
FIG.4
Run GLM-5.3 on Prime infrastructure and access more models through the Prime Inference Gateway.
“We worked with Prime Intellect to train Fast Ask on Lab — a small RL-trained subagent that helps the Ramp Sheets agent find answers inside spreadsheets. The result beat the frontier models on accuracy while running at faster speeds and a fraction of the cost. Rather than wait on a better frontier model, we trained our own for the workflow that mattered to us”
Karim Atiyeh
Ramp Co-CEO
“Evals are the foundation for building better agents. Prime Intellect helps turn them into real improvement loops.”
Robin Salimans
Principal AI Engineer
Environment Hub
Access and contribute to 2,500+ open-source RL environments and a community of researchers and developers.
Explore
My Stars
My Environments
primeintellectopencode-science
Solve science problems using OpenCode agent via...
Updated 8 days agoprimeintellectdeepdive
DeepDive QA RL environment with a Serper-powered search tool
Updated 11 days ago
stochi0rubric-discovery
Meta-environment for learning rubric functions from labeled...
Updated 2 months agoprimeintellectmini-swe-agent-plus
Mini SWE Agent Plus environment for solving SWE issues inside Pri...
Updated 3 days agoprimeintellectdeepdive
DeepDive QA RL environment with a Serper-powered search tool
Updated 11 days agoprimeintellectscience-env
A collection of challenging single-turn science problems
Updated 11 days agohud-text-2048
Text-based 2048 game for training agents to reach target tiles through strategic moves
Updated 7 months agohud-text-2048
Text-based 2048 game for training agents to reach target tiles through strategic moves
Updated 7 months agowill/tau2-bench
Verifiers implementation of tau2-bench
Updated 2 months ago
import verifiers as vf vf_env = vf.ToolEnv( dataset=dataset, parser=parser, rubric=rubric, tools=tool_list, max_turns=10, )
A library of modular components for creating RL environments and training LLM agents.
uv run rl \ --trainer @ examples/reverse_text/ rl/train.toml \ --orchestrator @ examples/ reverse_text/rl/orch.toml \ --inference @ examples/ reverse_text/rl/infer.toml
A framework for asynchronous reinforcement learning (RL) at scale.
deepswe-sandbox-1
python:3.11-slim
deepcoder-sandbox-1
python:3.11-slim
i3-math-sandbox-1
python:3.11-slim
For secure code execution optimized for large-scale reinforcement learning.
Inference. Run GLM-5.3 and other leading models
through one OpenAI-compatible API.
Prime Hosted Models run on Prime infrastructure. Prime Inference Gateway connects you to models served by third-party providers. Access both through the same API.
Find the right model for your workflow.
Evaluate models on your tasks, build with the API, and talk to us when you need dedicated serving capacity.
Compute. Find reliable compute operated globally from a single GPU to largest clusters.
On demand
Instant access to 1-256 GPUs.
Use your GPUs across clouds in a single platform.
FIG.5
H200
$1.99/HR
H200
$1.80/HR
H200
$1.23/HR
H200
$0.47/HR
B300
$4.99/HR
B200
$3.49/hr
H200
$3.14/HR
H100
$2.43/HR
Spot 0.94/HR
GH200
$3.14/HR
RTX Pro 6000
$3.14/HR
A100
$3.14/HR
A40
$3.14/HR
Liquid Reserved Clusters
Request large-scale clusters from 50+ providers.
Sell-back idle GPUs to our spot market.
FIG.7
Enter GPU name..
B300 SXM6 x 512
3-YEAR RESERVED$5.00/HR/GPU
TOTAL $2,560/hr
Profit on idle capacity
+$20,183,040
Research. Our Contributions to the Frontier of Open-Source AI
DISCOVER
Research
Prime Agent: A self-improving RLM agent
A self-improving agent harness built around RLMs and the Continual Harness.
Recursive Language Models: the paradigm of 2026
How we plan to manage extremely long contexts.
INTELLECT-3: A 100B+ MoE trained with large-scale RL
A 100B+ parameter Mixture-of-Experts model trained on our RL stack.
SYNTHETIC-2 Release
Four million collaboratively generated reasoning traces.
Join Prime Intellect
We are seeking the most ambitious developers to join our team — in San Francisco or remotely. Please send us examples of your exceptional work.