Models. Agents. Real work.

Build software.Get work done.

Corepass is a full-stack AI company building models, agents and a developer platform. Start with Remy to work through coding tasks and review the results.

For individualsFor developersFor teams

Core products & technology

Models and agents, built for real work.

Work with Remy, choose a model, or build your own agent experience.

01

An AI agent for coding and shipping.

Build it. Ship with confidence.
Remy turns your ideas into code and checks its own work.

Explore Remy

Describe it. Bring the spec and the design.

02Corepass Models

Advancing the frontiers of intelligence.

From lightweight specialists to flagship reasoning, we’re advancing model capability, efficiency, and reliability through research grounded in real-world tasks.

Scout
Lightweight specialists

Training compact models for well-defined, verifiable tasks, using specialization to improve the efficiency of agent workflows.

  • Code search
  • Verifiable tasks
  • Compact
Horizon
Flagship reasoning · In progress

Exploring deeper reasoning, coherent decision-making, and reliable task completion for complex engineering and long-horizon work.

  • Cross-module
  • Debugging
  • Long tasks
Voyager
Balanced capability · In progress

Balancing capability, speed, and cost for everyday development and frequent tasks, bringing capable models into practical work efficiently.

  • Features
  • Testing
  • Debugging
03Corepass Platform

Build AI that gets work done.

From model calls to complete agent execution, bring Corepass models, tools, and runtime environments into your own products—building AI applications that understand tasks, take action, and deliver results.

  • Horizon
  • Voyager
  • Scout
  • Agent Harness: Planning, context, and task execution.
  • Model API:
  • Agent API:
  • Tools & Runtime:
Bring models into your product

Integrate Corepass models into your applications, agents, or development frameworks, with control over context, tool calls, and interaction flows.

{/;>

Performance & efficiency

Measure the work.
Understand the cost.

Explore model and agent performance, with the tasks, run settings, and costs behind the results.

Model performance

Scout

A 4B code-exploration model, measured on all 848 tasks of SWE-Explore-Bench.

Retrieval F10.1802.3× the base model (0.077)
Time per task5.4s−29% vs the base model (7.6 s)

Industry solutions

Bring AI into the work
of your business.

LendingOS runs the lending workflow end to end, with underwriting and sales built in. Wavus finds and enriches the prospects who feed it.

01Lending operations

Bring applications, documents, and lending workflows into one place.

Explore LendingOS

Review loan documents against underwriting criteria, surface risks, and trace the reasoning behind each finding.

Keep contacts, follow-ups, activities, and deal context together in one sales workspace.

02Prospect intelligence

Find and enrich prospect information through natural-language search, with sources you can check.

Explore Wavus ↗

Trusted by

  • Cityverse
  • Flip Fundr
  • YouLand
  • PML
  • JSS Financial
  • Home Equity Advisor
  • Alameda
  • Attune
  • NexFlip
  • Cityverse
  • Flip Fundr
  • YouLand
  • PML
  • JSS Financial
  • Home Equity Advisor
  • Alameda
  • Attune
  • NexFlip

Start with your next task.Get to work with Remy, or build with