Full Breakdown
TypeSafe AI Secures $870 Million Series A and Expands “Jev” Judgment Model
By Drooid · · How we work
Core Funding Event
On October 9 2026, San Francisco-based TypeSafe AI announced a Series A financing that raised $870 million, valuing the company at $7.5 billion post-money. The round was led by Andreessen Horowitz, with participation from Sequoia Capital, DCVC and angel investors. Andreessen Horowitz partner Martin Casado joined TypeSafe’s board.
Background & Context
Founded in 2024 by Diogo Almeida, a former OpenAI researcher, TypeSafe released Jev on September 15 2026 as a “System One Model.” Unlike token-by-token LLMs, Jev returns pre-typed, probabilistic judgments (Choice, Score, Noul) that software can consume directly. The architecture uses a parallel sampler and a training method called Reinforcement Learning for Calibrated Decisions (RLCD).
Timeline
- Sept 15 2026 – Jev launched in early access.
- Oct 7 2026 – Case study: talent-marketplace Jack & Jill replaced Gemini 3.1 Flash Lite with Jev for candidate-matching.
- Oct 9 2026 – Series A closed; valuation and board changes announced.
Data & Statistics
- Financing: $870 million; $7.5 billion valuation.
- Pricing: $0.042 per 1 million input tokens; output tokens free.
- Performance: 70 – 500 ms latency; Jev claimed 193.6× faster and 444.6× cheaper than frontier models in internal tests.
- Adoption: TypeSafe says roughly one-third of Fortune 500 companies use Jev; a16z cites 25 % integration.
- Jack & Jill case study: Cost per 1,000 candidates fell from $0.755 to $0.092 (88 % reduction); median screening time dropped from 20.3 s to 10.3 s; selection rate stayed at 94.6 % vs. 93.9 % baseline. Reported annual savings $265,000, projected $500,000 over the next year.
Official Statements & Responses
- Diogo Almeida said Jev delivers “type-safe structured values” that eliminate hallucinations and type errors.
- Martin Casado noted a16z’s board seat will help guide the roadmap and highlighted “judgment-oriented AI” as a market segment.
- The company blog outlined plans for more machine-native models, expanded infrastructure, and enterprise-grade features.
Impact & Implications
Jev’s structured-output approach may reduce downstream text parsing and lower compute costs for enterprises. If speed and cost gains hold across workloads, judgment-focused models could become preferred for real-time classification, routing and automated decision pipelines. The large Series A underscores investor confidence in specialized AI architectures.
What’s Next
TypeSafe will extend Jev’s capabilities, release additional system-one models, and roll out the announced enterprise features. No specific future dates were disclosed.