Magic
Software engineers and developers seeking AI assistance for writing, reviewing, debugging, and planning code changes.
Overview
Magic is an AI company focused on building frontier code models to automate software engineering and research. The company's stated mission is to pursue safe AGI by automating AI research and code generation, combining frontier-scale pre-training, domain-specific reinforcement learning, ultra-long context, and inference-time compute. Magic has raised $515 million from investors including Nat Friedman, Daniel Gross, CapitalG, Elad Gil, Sequoia, Jane Street, and Eric Schmidt, and operates thousands of GB200s. Founded in 2022 and headquartered in San Francisco, the company is privately held with a team of 51-200 employees.
What it does
Magic develops advanced AI software engineers capable of handling multi-million token context windows. The company builds and trains large language models specialized for code generation and software engineering tasks, with a focus on ultra-long context and domain-specific reinforcement learning. It also conducts research on AI safety and AGI readiness, publishing policies and updates on its approach.
Key features
- Multi-million token context window
- Frontier-scale pre-training
- Domain-specific reinforcement learning
- Ultra-long context models
- Inference-time compute
- AI safety and AGI readiness policies
Use cases
- Automating software engineering tasks
- Code generation and review
- Debugging and planning code changes
- AI research and model improvement
- Handling large codebases with long context
Pricing
No pricing information is published on the company's website.
Pricing is gathered from the company's public pages and may change. Check the vendor's site before buying.
Pros and cons
Strengths
- Substantial funding ($515 million) from notable investors
- Advanced ultra-long context capabilities (100M token context windows)
- Strong focus on AI safety and AGI readiness
- Partnership with Google Cloud
- Backed by significant compute resources (thousands of GB200s)
Limitations
- No public pricing information available
- No self-serve or free tier mentioned
- Early-stage product with limited public documentation
- Primarily research-focused, may not be ready for mainstream enterprise adoption
What sets it apart
- Ultra-long context window (100M tokens)
- Combination of frontier-scale pre-training and domain-specific RL
- Explicit focus on safe AGI and AI readiness
- Partnership with Google Cloud
Frequently asked questions
What is Magic's main product?
Magic builds frontier code models to automate software engineering and research, with a focus on ultra-long context and domain-specific reinforcement learning.
How much funding has Magic raised?
Magic has raised $515 million from investors including Nat Friedman, Daniel Gross, CapitalG, Elad Gil, Sequoia, Jane Street, and Eric Schmidt.
Where is Magic headquartered?
Magic is headquartered in San Francisco, California.
When was Magic founded?
Magic was founded in 2022.
Does Magic offer a free tier or trial?
No free tier or trial is mentioned on the company's website.
What is Magic's approach to AI safety?
Magic has an AGI Readiness Policy that describes how it evaluates, monitors, and reduces existential risks of AI capabilities.
Sources
This profile was compiled from the company's own pages and public web research.
Last researched August 11, 2026.
Manage this listing
Own this company? Update your listing information.