TypeSafe AI, the developer of the “non-text” artificial intelligence model Jev, has reached a $7.5 billion valuation following an $870 million Series A funding round. Led by Andreessen Horowitz (a16z), the investment comes just 24 days after the model’s public launch on September 15, 2026.
The funding round also saw participation from Sequoia Capital and DCVC. The valuation marks a rapid ascent for the startup, which had previously been valued at $200 million during its seed stage. This capital injection is intended to scale the infrastructure for Jev, a model that deviates from the industry-standard Large Language Model (LLM) architecture by refusing to generate prose or conversational text.
Unlike OpenAI’s GPT-4 or Anthropic’s Claude, which are designed to predict the next word in a sequence, Jev is categorized as a “Decision Model.” It utilizes what TypeSafe AI describes as a “System One” architecture, optimized for high-speed, structured outputs rather than human-readable narrative. Instead of sentences, the model returns three specific primitives: Choice (selecting from a defined list), Score (a probabilistic value), and Noul (a boolean or null-state logic).
The technical shift toward non-text AI addresses a specific bottleneck in enterprise automation. While LLMs often struggle with latency, high operational costs, and “hallucinations” in structured data environments, Jev is designed to interface directly with software APIs. By outputting raw probabilistic values, the model can integrate into automated workflows—such as financial fraud detection or supply chain logistics—without the overhead of natural language processing.
TypeSafe AI is led by CEO Diogo Almeida, a former researcher at OpenAI who co-developed Reinforcement Learning from Human Feedback (RLHF). Almeida’s background in refining how models prioritize specific outcomes has been central to Jev’s design, which emphasizes reliability in decision-making over conversational fluency.
Reports indicate that approximately one-third of Fortune 500 companies have already begun adopting Jev since its mid-September early access release. This rapid adoption is attributed to the model’s design, which outputs structured probabilistic decisions instead of prose or code.
TypeSafe AI’s singular focus on non-text primitives remains a distinct approach in a market currently dominated by generative chat interfaces.
