Inside Multiplet

# Give your coding agent   a fleet.

256 agents with their own computers, model inference and a way to work together. Multiplet runs the fleet; you choose the project.

01 / Working agents

## 256 coding agents.  A computer for each.

Give the fleet your project, dependencies and objective. Agents can build software, run tests and try different implementations at the same time.

Multiplet provides model inference, the agent harness and the working environments as one service.
[Your code and tools, ready for the fleet ](/features/environments/) 

 agent-042 / workspace Linux 

project /src / tests / bench / results / 
A working environment 
Read the code.  Try a change.  Run the checks.

$ ./run-tests  Checking the implementation… 
Code · dependencies · tools · files

02 / Shared discoveries

## One finding can  move the fleet forward.

A useful test, a faster implementation or a new tool can spread through the fleet. Agents exchange messages and files, ask peers for help and build on each other’s work.

The division of work can emerge as they investigate. You don’t have to script every role or exchange.
[See how agents work together ](/features/coordination/) 
An example exchange 
Agent 014 → 042 
I found a failing case. Test saved in shared/repro.py.

Agent 042 → 086 
The fix passes here. Can you check it on the other platform?

Agent 086 → team 
Confirmed. I added the regression test.

03 / Back to your project

## Keep the code.  Use the work.

Changes, tests, experiments and findings return as files your existing agent can read. Ask it to explain what improved, run the checks and prepare the changes you want to keep.
[What comes back ](/features/artifacts/) 

Returned to your project run / 
01 Code & changesReview and apply
02 Tests & experimentsRun the checks again
03 Findings & activityFollow the investigation
Your agent helps you work through it.

Models 

### Leading open-weight models.

GLM, MiMo and DeepSeek, served together with the tools agents use to do the work.
[Models and tools ](/features/models-and-tools/) 
Coming soon 

### Steer the fleet as it works.

Add context, follow an approach and bring your judgment into the run.
[Live interaction ](/features/observability/) 

## Put a fleet on your project.

256 agents. Four hours. $1,000 for your first run, with model inference, computers and fleet operation included.

[Explore your first fleet ](/first-fleet/) [Ask your agent what to run ](/evaluate/)