TypeSafe raises $870M at $7.5B weeks after Jev launch

TypeSafe AI raised $870 million at a $7.5 billion valuation and says Jev passed $100 million ARR about three weeks after its September 15 launch.

TechCrunch reports that TypeSafe AI raised $870 million at a $7.5 billion valuation in a round led by Andreessen Horowitz, with Sequoia and existing investor DCVC participating. Jev, a non-text model that uses a transformer to output calibrated probabilities rather than text, launched on September 15, 2026; TechCrunch describes the financing as weeks later, while Latent Space says the $7.5 billion valuation came about three weeks after launch. Latent Space adds that TypeSafe said the Jev decision API passed $100 million in annual recurring revenue, with Sequoia cited on that figure, a claim TechCrunch does not report. The company says a third of Fortune 500 firms already use Jev and that it is faster and uses fewer tokens for automation. Co-founders are former OpenAI researcher Diogo Almeida, former Meta engineer Sasha Sheng, and Erik Gafni; the company was founded in 2024. Latent Space also lists same-cycle rivals—OpenAI’s Decisions API on GPT-6 Luna at $0.10 per million input tokens, Microsoft-Decision-1, Perplexity’s 27B pplx-decider (claimed 94.5% accuracy on 1,071 Decision Bench cases), and Cloudflare clef—and notes Claude workflows of up to 1,000 agents that benchmarks found often cost 1.8–5.1 times more for limited gains.

• TypeSafe AI raised $870 million at a $7.5 billion valuation, led by Andreessen Horowitz, with Sequoia and existing investor DCVC participating.
• Jev launched on September 15, 2026; TechCrunch says the round came weeks later, and Latent Space says about three weeks later.
• Latent Space says TypeSafe reported the Jev decision API passed $100 million ARR, with Sequoia cited; TechCrunch does not mention ARR.
• TypeSafe says one-third of Fortune 500 firms already use Jev, which outputs calibrated probabilities rather than text.
• Founded in 2024; co-founders are former OpenAI researcher Diogo Almeida, former Meta engineer Sasha Sheng, and Erik Gafni.
• Same-cycle rivals cited by Latent Space include OpenAI’s Decisions API on GPT-6 Luna at $0.10 per million input tokens, Microsoft-Decision-1, Perplexity’s 27B pplx-decider, and Cloudflare clef.
• Perplexity’s pplx-decider claims 94.5% Decision Bench accuracy across 1,071 cases.
• Latent Space also notes Claude workflows of up to 1,000 agents, with team runs costing 1.8–5.1 times more for limited gains.