Magic
Frontier code-model developer for autonomous software engineering and research.
Available information varies by company and source.
Profile record updated:
Company facts
- Official name
- Magic AI, Inc.
- Entity type
- COMPANY
- Founded
- 2022
- Headquarters
- United States
- Company size
- 50–200
- Market role
- B2B SaaS Provider
- Official website
- magic.dev
What Magic does
Magic builds proprietary coding-focused foundation models and deploys them through developer-facing software, applications and IDE extensions. It creates value by applying long-context model inference to large code repositories and complex engineering tasks, reducing manual software-development work. Its disclosed financing supports model training, research and compute capacity; customer pricing and revenue recognition are not disclosed.
Category differentiation
Magic is a frontier code-model and AI software-engineering company, not a general-purpose consumer AI assistant or a media/adtech platform. It develops proprietary coding models rather than operating solely as an interface over third-party foundation models.
Strategic context
AI-supported assessment from the existing company research; distinguish interpretation from sourced facts.
Magic AI, Inc. is a United States-based private AI software company developing proprietary frontier code models and software-engineering systems. Its technology addresses repository-scale code analysis, long-context inference, code generation, completion and autonomous engineering workflows for developers, engineering teams, technology companies and research organisations. Magic has raised substantial external capital to fund model research and compute infrastructure, including thousands of GB200 systems. The company’s commercial pricing and recognised revenue model are not disclosed in the supplied information. Its developer-facing software, desktop/mobile applications and IDE extensions establish a B2B product route to market, while the latest market signal states that it has not yet released a model.
Company news briefing
Briefing updated:
AI startup Magic has reported a significant breakthrough in pretraining efficiency, claiming its new recipe matches DeepSeek V4 Pro's base model quality using 50x less compute at an approximate cost of $500,000 on Nvidia GB200 systems. Having raised over $460 million—including a $320 million funding round valuing the enterprise at $1.5 billion—the company has partnered with Google Cloud for tens of thousands of GB200 chips, though its performance claims remain self-reported and unreleased to the public.
Business model & monetisation
Magic’s commercial monetisation mechanism is not disclosed. The supplied product data identifies a SaaS-platform delivery mode with developer applications and IDE extensions, but provides no subscription tiers, usage prices, licences, enterprise contracts, take rates or reported revenue. External equity funding has financed research and infrastructure rather than constituting operating revenue.
- Developer software and IDE-extension access
- Commercial terms not disclosed
- Model-powered autonomous engineering workflows
- Commercial terms not disclosed
Products & capabilities
No products with linked sources are available in this view.
Products & market categories
Technology
Recent recorded signals
Dates refer to the source publication. Older entries are historical context, not evidence of a new event.
Magic AI Claims Frontier-Level Pretraining for Under $1M
AI · Recorded impact score: 4/5
Magic, an AI startup co-founded by Austrians Eric Steinberger and Sebastian De Ro, claims a major breakthrough in pretraining efficiency. Its new recipe reportedly matches DeepSeek V4 Pro's base model quality using 50x less compute, costing about $500,000 on Nvidia GB200 systems, and is over ten times more compute-efficient than leading open-weight models. Scaling to roughly $4 million, Magic says it outperforms all public base models on perplexity evaluations, comparing against DeepSeek V4 Pro, Kimi K2, and Nvidia's Nemotron 3 Ultra, while excluding closed models from Anthropic, Google, and OpenAI. The company has raised over $460 million, with a $320 million round valuing it at $1.5 billion, and partners with Google Cloud for tens of thousands of GB200 chips. Magic has not yet released a model, and all claims are self-reported.
- Magic claims its pretraining recipe matches DeepSeek V4 Pro's quality with 50x less compute, costing ~$0.5M on GB200, and is 10x more efficient than leading open models.
- Scaling the recipe to ~$4M, Magic says it beats all public base models on perplexity evaluations.
Explore company relationships
Questions about Magic
What is Magic?
Magic is a private AI company that builds proprietary frontier code models and developer software for automated software engineering and research.
Who uses Magic?
Magic targets software developers, engineering teams, technology companies and organisations requiring automated engineering capabilities.
How does Magic make money?
Magic’s pricing and operating revenue model are not disclosed; its stated delivery route is B2B developer software and IDE extensions.
Sources & coverage
This profile uses public, official and technically observable information. Missing information does not prove that a product or relationship does not exist. The list below does not imply that every profile statement has been verified.
15 publicly documented primary sources and citations linked across the market graph.
Continue your research on Magic
Explorer includes additional company details, a Watchlist for up to 25 companies and your personal Strategic Intelligence Agent. It monitors your market daily and delivers tailored briefings with clear strategic context whenever relevant news occurs.
Free, with no time limit.