Fundraising Fox

Odyssey

Palo Alto, US Β· Founded 2023 Β· 74 employees on LinkedIn Β· 10 known investors

Odyssey is an AI lab developing general world models trained on multimodal observations of the physical world. The company builds foundational AI systems including Odyssey-2, Starchild-1, and Agora-1 for advancing physical simulation and multi-agent interaction.

Also known as Odyssey ML

Founders & leadership

Odyssey was founded in 2023 by Jeff Hawke and Oliver Cameron.

JH
Jeff HawkeCTO
OC
Oliver CameronFounder

Investors Β· 10

Also in the syndicate Β· 1

Natural Capitallead

Funding

SEC filings, press & company announcements

$310M disclosed across 1 of 2 rounds Β· 2025–2026

Source: company announcements and press reports β€” follow each round's link for the claim.

Company profile

researched Aug 2026

Odyssey is an artificial intelligence research lab developing world models β€” AI systems intended to understand and simulate the world rather than only generate or process language. The company positions world models as a successor category to language models and a foundation-model layer applicable to robotics, science, healthcare, education, gaming and energy.

Its published model family includes Odyssey-2, a general-purpose world model that takes an image or text prompt and produces long-running interactive simulations, streaming frame by frame with individual frames generated in roughly 50 milliseconds and simulations running for minutes rather than fixed short clips; Starchild-1, which extends beyond visually supervised world models toward learning from richer multimodal interaction; Agora-1, a multi-agent world model allowing several human or AI participants to share and interact within a single simulation in real time; and PROWL-1, a reinforcement-learning adversarial framework in which an agent explores game environments to improve world model performance. The company describes Odyssey-2 as trained on decades of video and as early-stage technology.

Odyssey frames three broad user groups for its models: humans, for educational and entertainment experiences; AI systems, for simulated environments that challenge agents; and robots, for simulating actions before executing them. It also outlines prospective near-term applications such as interactive gaming, robotic intelligence, tutoring, retail and hospitality guidance, healthcare navigation, virtual try-on, interactive advertising and employee training.

Business model

Odyssey offers access to its world models through an API, with sign-up for API access and interactive demonstrations of Odyssey-2 and Agora-1 available on its website.

Traction

Publicly stated traction consists of the announced $310M Series B, availability of interactive trials for Odyssey-2 and Agora-1, and published technical reports and research papers.

Latest developments

Odyssey has announced a $310M Series B and publicizes four systems β€” Odyssey-2, Starchild-1, Agora-1 and PROWL-1 β€” with accompanying technical reports and papers.

β–ΈFull profile β€” market position, technology, go-to-market

Market position

Odyssey presents itself as a lab pioneering world models as the next foundation-model category, differentiating its approach from language models and from conventional short-clip video generation models.

Odyssey emphasizes real-time, open-ended interactivity and duration: simulations begin instantly, stream for minutes, and accept ongoing interaction, versus fixed-length, non-interactive outputs from traditional video generation systems. It also claims material advances in the physical accuracy of world models.

Technology

The core technology is generative world models that simulate environments frame by frame in real time. Odyssey-2 accepts image or text prompts and streams interactive, minutes-long simulations with roughly 50 ms per-frame generation, contrasted by the company with traditional video models that produce fixed 10-second, non-interactive clips after minutes of generation. Starchild-1 targets learning from multimodal interaction beyond visual observation, Agora-1 supports shared real-time multi-agent simulation, and PROWL-1 applies reinforcement learning agents adversarially within game environments to improve world model quality.

Go-to-market

Distribution is direct via the company's website, which offers API access requests, model trial pages, technical reports and research papers, plus a contact and careers presence.

The company describes its models as serving humans, AI systems and robots, with prospective use cases across gaming, robotics, education, healthcare, retail, hospitality, advertising, fitness and enterprise training.

Compiled by commissioned research from 1 cited public sources β€” announcements, filings, and press listed under research sources below.

Key figures

latest reported
HeadcountAug 202674
Odyssey-2 frame generation latencyJan 202550 milliseconds

Company-reported or press-reported figures, each dated to when it was claimed β€” not independently audited.

Competitors Β· 3

by search overlap

Companies competing with Odyssey for the same Google search keywords, organic and paid, via search-intersection analysis.

Timeline Β· 5

launches, deals, and filings
Jan 2025
Odyssey announces $310M Series B

Odyssey states on its website that it raised a $310M Series B.

$310M source β†—

Jan 2025
Agora-1 multi-agent world model

Agora-1 allows multiple participants, human or AI, to share and interact within the same world simulation in real time.

source β†—

Jan 2025
Odyssey-2 general-purpose world model

Odyssey-2 is presented as the company's most powerful general-purpose world model, generating interactive, multi-minute simulations from an image or text prompt with frames produced in about 50 ms.

source β†—

Jan 2025
PROWL-1 reinforcement learning framework

PROWL-1 is an RL-driven adversarial framework in which an RL agent explores game environments with the objective of improving world model performance; an accompanying paper is published.

source β†—

Jan 2025
Starchild-1 multimodal world model

Starchild-1 is described as moving beyond world models that learn only from visual observation toward systems that learn from richer multimodal interaction with the world; a technical report is published.

source β†—

Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.

In the news

β–ΈResearch sources Β· 1

primary sources listed

1 public sources were cited for this profile; the first-party ones are listed here.

Frequently asked questions

What does Odyssey do?
Odyssey is an AI lab building general world models that understand and simulate the physical world in real time.
Who founded Odyssey?
Odyssey was founded by Jeff Hawke, Oliver Cameron in 2023.
Who are Odyssey's investors?
Odyssey's investors include Air Street Capital, Evolution VC Partners, In-Q-Tel, MVP Ventures, Point Nine, Rocketship.vc, Thursday Ventures, Union Square Ventures (USV) and 1 more.
How much funding has Odyssey raised?
Odyssey has disclosed $310M raised across 1 of its 2 known rounds.
Where is Odyssey headquartered?
Odyssey is headquartered in Palo Alto, US.