TypeSafe AI raises $870M Series A at $7.5B valuation to deliver more AI models – Unite.AI

TypeSafe AI announced on October 9, 2026 that it raised an $870 million Series A at a $7.5 billion valuation, led by Andreessen Horowitz with participation from Sequoia Capital, existing DCVC investor and angel investor. Andreessen Horowitz general partner Martin Casado joins the firm’s board of directors.

Round terms and investors

Andreessen Horowitz confirmed it is leading the investment in a post dated October 9, 2026 and signed by Jennifer Li, Sarah Wang, Martin Casado, Marc Andreessen and Ben Horowitz. Li and Casado are general partners focused on the firm’s infrastructure investments, Wang is a general partner on its growth team, and Andreessen and Horowitz are cofounders and general partners of the firm.

TypeSafe’s announcement is a short post aimed at developers, businesses, and potential recruits; the company disclosed the dollar amount, valuation and Casado’s position on the board of directors in a footnote to that post.

The System One model behind the round

TypeSafe released Jev into early access on September 15, 2026, after two years of stealth, and said it is removing developers from its waitlist. In the launch post, founder Diogo Almeida described Jev as the first entry into a new class of models the company calls System One, designed for rapid, structured decision-making within the software.

According to TypeSafe documentation, Jev takes typed questions and evaluates them against a provided state, returning structured results without generating or parsing text. The results arrive as values ​​typed with probability distributions, so your code can branch, sort, and route directly on them. The company exposes three primitives: Choice chooses an option from a provided list; The score evaluates the state against a rubric; and Noul returns a 0–1 answer to the question of whether a given statement is true. Choice and Score also return confidence values, and all three question types can be combined into a single API call, where each is evaluated independently and simultaneously against the same state.

Almeida wrote that TypeSafe has created a new stack focused entirely on automation: a new model architecture, a parallel sampler, and a training method that the company calls Reinforcement Learning for Calibrated Decisions (RLCD). He said Jev achieves similar intelligence to existing large language models on System One tasks while executing two orders of magnitude faster, with end-to-end response times of 70 to 500 milliseconds, a range that, according to the post, can be 40 to 200 times faster than frontier models with comparable intelligence for System One-shaped queries. Because the outputs are type-independent structured values ​​defined in advance, Almeida wrote, the model never makes type errors and cannot hallucinate.

The published price is USD 0.042 per million input tokens, while output tokens are free. The company says the 193.6x faster and 444.6x cheaper figures on its homepage come from workflow evaluations, and warns that those numbers are on the higher end of real-world earnings. It also notes that the evaluation workflows were created by members of the model capabilities team and that some errors may exist.

Almeida wrote that at OpenAI he helped develop instruction-following methods for language models, work that he said became the research behind ChatGPT. The name System One refers to Daniel Kahneman’s book Thinking, Fast and Slow and its contrast between the quick, intuitive thinking of System 1 and the slower, more deliberate reasoning of System 2. Jev is named after William Stanley Jevons; the company said it expects AI to follow the path of coal, for which efficiency improvements in steam engines have led to increased demand.

Reported adoption and early deployment

TypeSafe said in its announcement that a third of Fortune 500 companies use Jev and that it has already saved customers millions of dollars in production. Andreessen Horowitz’s post states a different figure, stating that 25% of Fortune 500 companies have integrated Jev. The company described Jev as the fastest-growing model it has ever seen, saying that it reached 1 trillion tokens generated within three days of launch and that thousands of use cases emerged within the first week, including generative user interface, interactive gaming and data analytics. He characterized Jev as being 1/100th to 1/500th the cost of frontier models while being 100 times faster for classification tasks with comparable accuracy, passing model decisions directly into the code as typed values ​​rather than as text that the software must then parse.

In an October 7, 2026 case study, TypeSafe reported that Jack & Jill, a talent marketplace where AI agents match candidates with hiring teams, replaced Gemini 3.1 Flash Lite for 100% of calls at a key stage of its candidate matching pipeline within 10 days of its first test. According to the case study, the change reduced screening costs by 88%, from $0.755 to $0.092 per 1,000 candidates, and halved the average screening time from 20.3 seconds to 10.3 seconds, while retaining 94.6% of candidates that hiring managers subsequently asked to meet with, compared to 93.9% baseline; the study finds that this difference in quality was not statistically significant. Ranking quality was 0.933 for Jev versus 0.924 for Gemini on the AUC. TypeSafe reported $265,000 in annual savings for Jack & Jill at current volume, with $500,000 expected over the next twelve months given expected growth, and said Jev usage has expanded to more than 15 workflows, including matching roles shown to candidates when they sign up and a profile check that dropped from about 3.6 seconds to about 0.2 seconds.

Declared plans

TypeSafe said it intends to further develop Jev’s capabilities, provide additional machine-native models, provide infrastructure for building intelligent software, and add enterprise features requested by customers. The ad also encourages potential recruits to join by linking to the company’s careers page.