whereismyai
AI Scientist v3
AI Scientist v3 is an automated scientific discovery system that orchestrates coding agents to run experiments and write papers.
Agentic Scientist orchestrates coding agents to conduct automated scientific discovery. Starting from a task description, the system proposes ideas, runs experiments, evaluates outcomes, executes baselines and ablations, performs post-hoc analyses, and writes a complete research paper. It supports multiple AI agents and models including Claude, Codex, Cursor, Opencode, Gemini, and Pi. The pipeline runs securely inside Docker containers to handle arbitrary commands and API keys. Researchers can also inspect runs through an included web interface and send manuscripts to an external reviewer.
AI Scientist v3 at a glance
- Best for: Researcher, Software engineer
- Runs on: Docker, CLI, self-hosted, open source
- Pricing: Free
- Released: October 2026
- Category: AI research tool
- Website: github.com/sakanaai/ai-scientist-v3
Key features
- Run automated multi-phase research pipelines
- Execute code inside isolated Docker containers
- Support multiple coding agents and LLM models
- Perform automated baselines and ablation studies
- Generate complete compiled PDF research papers
- Inspect experiment runs via a built-in web UI
- Review manuscripts with automated external reviewers
Use it for
- Automating scientific discovery and experimentation
- Running baseline comparisons and ablations
- Drafting task evaluators from research briefs
- Generating and compiling academic research papers
Alternatives to AI Scientist v3
- Paper2Agent — It enables direct interaction with research papers and automatic result reproduction.
- Auto Deep Researcher — Automates experiment operations to reduce manual babysitting of training loops.
- OpenResearch — Run parallel research workflows locally or on remote compute.
- AutoResearch — It provides a stateful, reproducible workflow for automated AI research.
- FAROS — Connects literature evidence, executable plans, and human review into an auditable research loop.
- DiscoveryRSI — Achieves high actionable-reference rates and supports multi-agent evolutionary research search.
- Arbor — It grows a hypothesis tree and tests changes on held-out data.
- SoL Pi — Saves $8.75 to $13.50 per hour compared to native Codex and Claude Code systems.
- Alook — Connects existing local coding agents to team communication channels.
- Transformer Lab — Accelerates the scientific research loop by executing experiments at high speed.
- AI Research Skills — 98 expert skills for training, evaluating and writing up ML research
- Jupyter AI — Connects frontier AI agents directly to JupyterLab notebooks.
All AI Scientist v3 alternatives