---
title: "A16z valued TypeSafe's Jev at $7.5 billion; Microsoft matched its $0.042 per million tokens"
description: "Andreessen Horowitz led an $870 million round valuing TypeSafe's decision model Jev at $7.5 billion on October 9, 2026. Two days earlier Liquid AI had made a rival free below $10 million of revenue, and the same day Microsoft shipped its own at Jev's exact price, built on Alibaba's open Qwen."
publisher: "The Inference"
section: "Money"
published: 2026-10-10T19:03:38.792Z
modified: 2026-10-10T19:03:38.792Z
canonical: https://theinference.org/article/a16z-valued-typesafe-s-jev-at-7-5-billion-microsoft-matched-its-0-042-per-million-tokens
language: en
keywords: "Venture Capital, AI Models, Pricing, Microsoft, Open Weight Models"
---

# A16z valued TypeSafe's Jev at $7.5 billion; Microsoft matched its $0.042 per million tokens

> Andreessen Horowitz led an $870 million round valuing TypeSafe's decision model Jev at $7.5 billion on October 9, 2026. Two days earlier Liquid AI had made a rival free below $10 million of revenue, and the same day Microsoft shipped its own at Jev's exact price, built on Alibaba's open Qwen.

## Two prices for one decision, set the same day

On October 9, 2026, Andreessen Horowitz led roughly $870 million into TypeSafe AI [at a $7.5 billion valuation, 24 days after a $40 million seed priced near $200 million](https://a16z.com/announcement/investing-in-typesafe-ai/) [1][2]. The same day, [Microsoft](https://theinference.org/markets/companies/microsoft) put Microsoft-Decision-1 [on its Foundry service at $0.042 per million input tokens](https://cryptobriefing.com/microsoft-launches-decision-1-ai-model/), output free [4][5]. Jev lists at exactly that price [6]. Microsoft did not undercut the startup; it matched it to the tenth of a cent.

*Chart: **From a $200 million seed to a $7.5 billion round in 24 days** Bar chart of TypeSafe AI valuations: a roughly $0.2 billion seed in September 2026, a $7.5 billion Series A in October 2026, and a talked-up next round above $10 billion marked as an estimate.*

*Post-money valuations for TypeSafe AI: a $40 million seed led by DCVC at about $200 million on September 15, 2026, and an approximately $870 million Series A led by Andreessen Horowitz at $7.5 billion on October 9, 2026. The third mark is reported investor interest in a round of more than $1 billion at a valuation above $10 billion, not a closed deal. [1][2]*

## Three weeks took the category from one model to a hundred

Jev passed a million users in days; [a16z](https://theinference.org/markets/companies/a16z) credits it with a trillion tokens in three days and integration at a quarter of the Fortune 500, a third by the startup's count [2][3]. [Cloudflare](https://theinference.org/markets/companies/cloudflare)'s open-weight Clef arrived on October 1 as [a drop-in for Jev's API](https://www.theregister.com/ai-and-ml/2026/10/01/cloudflare-tries-to-outplay-jev-with-open-weight-clef-models/5300649) [6], [OpenAI](https://theinference.org/markets/companies/openai)'s Decisions API hit public beta [7], and more than 100 such models are vying for attention [4]. On October 7, [Liquid AI](https://theinference.org/markets/companies/liquid-ai) [open-weighted d1-3B and d1-omni-600M](https://www.liquid.ai/blog/d1-open), free for commercial use below $10 million of revenue [8][9]. d1-3B self-scores 48.57 on the Decision Index, level with a model 12 times its size, though Liquid ran the scorer itself rather than submitting to the board [8][10].

## The biggest entrants all start from Alibaba's open weights

Microsoft [post-trained Decision-1 on Qwen3.5-9B](https://www.marktechpost.com/2026/10/09/microsoft-ai-releases-microsoft-decision-1-a-qwen3-5-9b-decision-scoring-model/), Cloudflare's Clef-flash sits on the same base, and H2O's Lightning-4B on Qwen3.5-4B [4][5][6]. Future versions will rebase on Microsoft's own MAI and OpenAI models [4][12]. The money has not pooled where the weights are: [DeepSeek kept about 8 percent of estimated OpenRouter spending on its own models](https://www.msn.com/en-xl/technology/artificial-intelligence/china-s-open-weight-ai-models-are-winning-global-users-who-is-capturing-the-value/ar-AA2dJZaz), and Z.ai's token share fell from 88 to 22 percent in 20 days once rivals hosted it, Epoch AI found [11]. [Alibaba](https://theinference.org/markets/companies/alibaba) supplies the base layer; the hosts set the price.

*Chart: **One decision, six prices: free to $0.24, with Microsoft matching Jev** Bar chart comparing decision model prices per million input tokens, from a free self-hosted license through Liquid and Jev and Microsoft at $0.042, OpenAI at $0.10 and Cloudflare at $0.24.*

*Posted prices per million input tokens as of October 9, 2026; decision models bill no output tokens. The self-hosted license is free below $10 million of annual revenue under LFM Open v1.0, with hardware and serving excluded. Vendor prices as reported by MarkTechPost, The Register and explainx. [5][6][9][13]*

## The round is priced at about 1,500 times launch-pace sales

A trillion tokens is $42,000. That is Jev's posted price on the volume a16z says it handled in three days [3][6]. Held for a year, that pace is roughly $5 million of posted-price revenue, a 1,500-times multiple on the $7.5 billion mark. The inputs are a16z's figure, a launch spike and a list price enterprise contracts may not pay; TypeSafe discloses no revenue [1]. The bet needs demand compounding faster than the price of a call can fall.

## Adoption is real, and switching is sticky

Jev still beats the open models on performance charts, Liquid's included [14], and TypeSafe keeps its architecture secret [6]. Calibration scores do not transfer between models, so switching vendors means retuning every threshold [7]. Microsoft's own benchmarks are vendor-run, its latency method disputed by H2O.ai and its model absent from the JevBench board [5]. A price matched by a hyperscaler is a ceiling, not a moat.

## What decides it

Either the calls stay a standalone layer and the talked-up round above $10 billion closes [1], or they become a bundled feature and the money pools with the hosts. The tell: whether that round signs, and whether Jev's $0.042 holds once Microsoft lands on OpenRouter beside it [5][12].

## Takeaway

A16z's $7.5 billion bet on TypeSafe's Jev is priced against a category that went from one model to a hundred in three weeks, with Microsoft matching its price to the tenth of a cent on Alibaba's open weights: the round closes or the money pools with the hosts.

## Sources

1. [Crypto Briefing, a16z leads TypeSafe AI's Series A at a $7.5 billion valuation, 9 October 2026](https://cryptobriefing.com/a16z-leads-typesafe-ai-series-a/)
2. [Bloomberg, Andreessen Horowitz Backs Jev Maker at $7.5 Billion Value, 9 October 2026](https://www.bloomberg.com/news/articles/2026-10-09/andreessen-horowitz-backs-jev-maker-at-7-5-billion-value)
3. [Andreessen Horowitz, Investing in TypeSafe AI, 9 October 2026](https://a16z.com/announcement/investing-in-typesafe-ai/)
4. [The Register, Microsoft leans on open weight model from Chinese AI lab to challenge Jev, 10 October 2026](https://www.theregister.com/ai-and-ml/2026/10/10/microsoft-leans-on-open-weight-model-from-chinese-ai-lab-to-challenge-jev/5302473)
5. [MarkTechPost, Microsoft AI Releases Microsoft-Decision-1: A Qwen3.5-9B Decision-Scoring Model, 9 October 2026](https://www.marktechpost.com/2026/10/09/microsoft-ai-releases-microsoft-decision-1-a-qwen3-5-9b-decision-scoring-model/)
6. [The Register, Cloudflare tries to outplay Jev with open-weight Clef models, 1 October 2026](https://www.theregister.com/ai-and-ml/2026/10/01/cloudflare-tries-to-outplay-jev-with-open-weight-clef-models/5300649)
7. [InfoWorld, As AI agents grow, vendors carve out decision-making as a separate model layer, 9 October 2026](https://www.infoworld.com/article/4233075/as-ai-agents-grow-vendors-carve-out-decision-making-as-a-separate-model-layer.html)
8. [Liquid AI, Open d1: Edge decision models for text, vision, and audio, 7 October 2026](https://www.liquid.ai/blog/d1-open)
9. [MarkTechPost, Liquid AI Releases Open-Weight d1-3B and d1-omni-600M: Multimodal Decision Models With Zero Output Tokens, 7 October 2026](https://www.marktechpost.com/2026/10/07/liquid-ai-releases-open-weight-d1-3b-and-d1-omni-600m-multimodal-decision-models-with-zero-output-tokens/)
10. [Hugging Face, LiquidAI/d1-3B model card, accessed 10 October 2026](https://huggingface.co/LiquidAI/d1-3B)
11. [South China Morning Post, China's open-weight AI models are winning global users. Who is capturing the value?, 7 October 2026](https://www.msn.com/en-xl/technology/artificial-intelligence/china-s-open-weight-ai-models-are-winning-global-users-who-is-capturing-the-value/ar-AA2dJZaz)
12. [Crypto Briefing, Microsoft launches Decision-1, an AI model built to make fast calls, 9 October 2026](https://cryptobriefing.com/microsoft-launches-decision-1-ai-model/)
13. [explainx.ai, Liquid AI d1 Vision: 19-200x Cheaper Than GPT-6.1 Sol, 6 October 2026](https://explainx.ai/blog/liquid-ai-d1-vision-decision-model-200x-cheaper-than-gpt-6-1-sol-2026)
14. [GIGAZINE, Two decision-making models have been released: d1-3B and d1-omni-600M, 8 October 2026](https://gigazine.net/gsc_news/en/20261008-d1-3b-omni-600m/)


---

A16z valued TypeSafe's Jev at $7.5 billion; Microsoft matched its $0.042 per million tokens — The Inference. Canonical: https://theinference.org/article/a16z-valued-typesafe-s-jev-at-7-5-billion-microsoft-matched-its-0-042-per-million-tokens