Two Delta, a secretive Israeli artificial intelligence startup founded in 2025, has raised tens of millions of dollars in a seed round backed by investors including Bill Ackman, founder and CEO of Pershing Square, and former Google chairman Eric Schmidt. The company has disclosed little about its business beyond research papers published on its website.
Two Delta was co‑founded by Dvir Zeltzer and Dr. Eli David. David is described as a veteran AI researcher and entrepreneur, while Zeltzer brings a background in business and management, including a senior role at Google. The company says it is developing technology to accelerate AI models and improve the efficiency with which they use computing resources.
Two Delta’s DCOCO research with Nvidia
In October 2026 Two Delta published results from a collaboration with Nvidia showing its DCOCO technology used to accelerate inference for Nvidia’s Nemotron 3.5 Super VL model. DCOCO is built on speculative decoding, an approach in which a smaller, faster model drafts a sequence of tokens that a larger model then verifies in a single pass. The technique can speed generation but risks producing proposed tokens that the larger model rejects; the challenge is to produce drafts the larger model is more likely to accept.
Two Delta’s method combines candidate‑token embeddings with contextual information from the drafting model’s hidden states, which are informed by the target model’s activations. It scores combinations of neighboring tokens in parallel to select more consistent sequences before the larger model verifies them.
In tests across 10 datasets covering mathematics, coding and general‑purpose tasks, Two Delta reported that DCOCO achieved average throughput 2.03 times that of standard autoregressive decoding when reasoning was disabled and greedy decoding was used. Under the same configuration it reported an average throughput improvement of 19.8% over the competing DFlash method. The company also reported throughput gains under different decoding settings and workloads involving up to 32 concurrent requests. Evaluations were conducted on Nvidia DGX B200 hardware using the vLLM inference framework. These results were reported by Two Delta rather than an independent assessment; the company says DCOCO improves token‑generation efficiency while preserving the target model’s output quality.
EvoWeb benchmark for AI coding agents
Two Delta has also published EvoWeb, a benchmark designed to evaluate how well AI coding agents build and maintain web applications as users repeatedly request new features and modifications. Unlike single‑prompt benchmarks, EvoWeb tests an application over 10 successive requests: an initial build followed by nine changes.
The study tested 32 web applications and found that none of the evaluated models maintained full functionality throughout the process. The top performers, Fable 5 and Kimi K3, received scores of 0.694 and 0.684, respectively, on the study’s zero‑to‑one scale. Two Delta identified integration of new features with existing functionality as the main difficulty: build failures remained rare, but runtime errors increased from 2% at the beginning of the process to 12% after 10 requests.
Two Delta’s fundraising and its published research together highlight the company’s focus on improving inference efficiency and evaluating long‑running software maintenance by AI agents.
