Back to tools
Byte Goose Intelligence

Pilot Goose Agent

Autoresearch on your machine

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Download for macOS View rankings

Coming soon: this download link is a placeholder and will not work until the macOS app is released.

How it runs

Two ways to run Pilot Goose Agent

Choose where your research agent and tools run.

Mode 1

External harness

Run on your machine or your own cloud using your preferred coding agent.

  • Claude Code
  • Codex
  • OpenCode
  • Byte Goose Code
  • Custom agent

Mode 2

Managed harness

Run in managed cloud infrastructure with a dedicated environment for your research.

  • Tenant container / VM
  • Agent runtime
  • Python / tools
  • Persistent storage

Deployment options are illustrative; Pilot Goose Agent is not available for download yet.

Research insights

AI model rankings for research

Model token usage in Pilot Goose Agent over the last 30 days, grouped by UTC date.

Example data — not actual usage

Total tokens

160M

Models

6

Model makers

4

Top models

Daily input and output tokens

Last 30 days · UTC

Model usage distribution

Tile size represents each model’s share of tokens. Colors indicate the model maker.

  • GPT-4o mini · OpenAI31.8%
  • Claude Sonnet · Anthropic23.5%
  • Gemini Flash · Google16.8%
  • GPT-4o · OpenAI12.5%
  • Llama · Meta8.5%
  • Claude Haiku · Anthropic6.8%

Model maker leaderboard

Ranked by total token usage

  1. 1OpenAI70.9M tokens
  2. 2Anthropic48.5M tokens
  3. 3Google26.9M tokens
  4. 4Meta13.6M tokens