Developer docs
API playgroundTry for free, no card

Search company profiles

Odyssey

Full company profile

uuid00007oc

Namestring
Odyssey
Legal namestring
Odyssey
Websiteurl
odyssey.world
Company typeenum
Private
Founded yearint
2023
Descriptiontext

Odyssey is a private AI research lab founded in 2023 by Oliver Cameron and Jeff Hawke, both veterans of the autonomous vehicle industry. The company develops general-purpose world models—causal, multimodal AI systems trained primarily on video data to simulate physical environments with accurate physics. Its product portfolio comprises four named systems: Odyssey-2, a flagship world model that generates long-running interactive video simulations from text and image prompts with 50ms frame latency; Starchild-1, a model that learns from richer multimodal interaction beyond visual observation; Agora-1, a multi-agent world model enabling real-time human and AI co-inhabited simulations; and PROWL, a reinforcement-learning-driven adversarial research framework. The company is headquartered in Palo Alto with additional offices in Los Angeles, London, and Zurich, and employs approximately 55 people as of June 2026.

The underlying technology is differentiated by frontier claims in physical accuracy, long-horizon interactivity (multi-minute streaming versus fixed 10-second clips), and a documented cost advantage of achieving 80% model FLOP utilization on AWS Trainium3 chips—approximately double the 40-50% industry average. Odyssey targets five horizontal verticals: robotics (primary), gaming and interactive entertainment, healthcare and life sciences, defense and security, and scientific research, with use cases spanning robot task rehearsal, procedural game-world generation, medical training simulation, defense scenario generation, and physics-based scientific modeling.

The business model is API-first with an enterprise sales overlay. Odyssey offers API access to its world models through its website (odyssey.ml) and distributes through a strategic preferred-cloud partnership with Amazon Web Services, which also invested in the company's $310 million Series B in June 2026 (led by Natural Capital, with Amazon, AMD Ventures, Google Ventures, EQT, and In-Q-Tel participating) at a $1.45 billion post-money valuation. Revenue mechanics and pricing tiers are not publicly disclosed, and no customer logos have been disclosed in available sources. The company also benefits from sovereign AI infrastructure access through the UK government's AI Research Resource (Isambard-AI and Dawn supercomputers) as part of a £500M national AI initiative.

Short descriptiontext

Odyssey is a private AI lab developing general-purpose world models—causal, multimodal systems that simulate physical environments with accurate physics—targeting robotics, gaming, healthcare, defense, and scientific research applications via API access and an AWS preferred-cloud partnership.

Operating statusenum
Operating
Ownership categoryenum
Headcount rangeband
51–100
akta.pro rankint
HeadquartersPalo Alto, United States
HQ citystring
Palo Alto
HQ countrystring
United States
HQ regionstring
North America
Markets served

Serves global market

Offices3 records

Each record includes

City, Country, Type, Description, Source

Keyword5 values
generative world models, AI simulation platform, physics-based video generation, multimodal AI models, developer API access
Industry2 codes
PRIMARYHDAEANAA
End-to-End Enterprise AI Platforms (MLOps & Model Lifecycle Management)
Hardware, Technology & Telecom/Enterprise Platforms & Databases/Artificial Intelligence Platforms for Enterprise
SECONDARYHDAEANAC
Model Serving, Inference & Deployment Platforms (APIs, Edge/On-Prem)
Hardware, Technology & Telecom/Enterprise Platforms & Databases/Artificial Intelligence Platforms for Enterprise
NAICS code1 code
  • Software Publishers513210
SIC code1 code
  • Services-Prepackaged Software7372
Product category
Generative AI / World Model Simulation
Social media profiles3 records
GTM motion2 records

Each record includes

Type, Description, Source

Revenue model1 record
1API Access / Model Licensing
TypeSubscription Recurring
Description

Odyssey provides API access to its world model technology, suggesting a usage-based or subscription licensing model for developers and enterprises integrating world simulation capabilities.

thetechportal.com
Marketing channels3 records

Each record includes

Title, Type, Stage, Description, Source

Distribution channels2 records

Each record includes

Title, Type, Scope, Target buyer, Description, Source

Cost components4 values
Technology or R&D, Infrastructure, Personnel, Marketing or Sales
GTM typeB2B
B2B
Offering typeSoftware
Software
Brand1 of 4 records shown
1Odyssey-2
Description

Most powerful general purpose world model that materially advances state-of-the-art in physical accuracy of world models

odyssey.ml
+3 more records
Core offering1 text field

Odyssey is an AI lab that develops general-purpose world models—causal, multimodal AI systems that learn from video data to predict and simulate physical environments with accurate physics. Its flagship Odyssey-2 model generates long-running, interactive video simulations from text and image prompts, while complementary models (Starchild-1, Agora-1, PROWL) extend capabilities into multimodal learning, multi-agent simulation, and reinforcement-learning-driven adversarial training. The technology is delivered to developers and enterprises via API access and through a strategic AWS cloud partnership.

Differentiator
Functional benefit
Problem solved
Quantifiable outcome1 of 3 values shown
  • 80% model flop utilization on Trainium3 chips (vs 40-50% industry average)
+2 more records
Product overview1 text field

Odyssey is an AI lab developing general-purpose world models—causal, multimodal systems that learn to predict and interact with the physical world over long horizons. The product portfolio centers on world model technology, with four key offerings: Odyssey-2 (flagship general-purpose world model generating interactive video simulations), Starchild-1 (multimodal world model learning from richer interaction), Agora-1 (multi-agent world model for real-time collaborative simulation), and PROWL (reinforcement learning research framework). These products form an integrated stack enabling applications in robotics, gaming, healthcare, and scientific simulation.

Product and service1 record
1Odyssey-2
Scale indicator6 records

Each record includes

Type, Value, Description, Source

Partnership1 partner
Strategic tierCoreTypeTechnology or IntegrationAnnounced on2026-06-17
Description

Strategic partnership making AWS Odyssey's preferred cloud provider. Odyssey will use Amazon Trainium chips for compute-intensive world model training workloads. This is a key strategic relationship for both companies.

Recent move6 records

Each record includes

Date, Type, Title, Description, Source

Expansion highlight8 records

Each record includes

Type, Description

Peers10 records
TypeEmerging player
Description

Pika develops generative video models with an emphasis on controllable, interactive video creation. It shares Odyssey's modality (generative video) and a portion of its target use cases (creative content, simulation-style previews), though Pika is more focused on creative tooling than physical-world simulation.

TypeDirect peer
Description

World Labs is a direct competitor developing foundation world models for spatial intelligence and 3D scene generation, with the same core thesis of generative world models as Odyssey. Both target enterprise applications including robotics, VFX, and simulation with API-accessible models.

TypeBroad incumbent
Description

Stability AI is a foundation-model company offering generative models across images, video, and audio via APIs. It competes with Odyssey in the broader generative AI infrastructure space and overlaps in serving developer and enterprise customers with multimodal generation capabilities.

TypeBroad incumbent
Description

Google DeepMind develops Genie and other foundation world models as part of its broader AI research portfolio. While not focused exclusively on world models like Odyssey, DeepMind's scale, compute access, and multimodal research make it the most consequential incumbent competitor in the same technology category.

TypeBroad incumbent
Description

NVIDIA's Cosmos platform provides world foundation models for physical AI and robotics simulation, overlapping directly with Odyssey's robotics and simulation use cases. As a broader AI infrastructure incumbent with deep compute partnerships, NVIDIA can both compete with and enable Odyssey's go-to-market.

TypeDirect peer
Description

Decart builds real-time generative world models and interactive video environments, mirroring Odyssey's focus on long-horizon, interactive simulation. The companies share a target customer base of developers and enterprises integrating world-model APIs into robotics, gaming, and creative tools.

TypeEmerging player
Description

MiniMax's Hailuo AI offers high-quality generative video models with a developer and consumer-facing API, competing in the same multimodal generative video category as Odyssey's world models. It is an emerging player with overlapping capabilities in text/image-to-video generation.

TypeEmerging player
Description

Lightricks offers AI-powered video generation tools (LTX Studio, etc.) targeting creators and marketers. It is a smaller, emerging player with partial overlap to Odyssey in generative video and interactive media, though focused more on creative workflows than physics-accurate world simulation.

TypeEmerging player
Description

Runway offers generative video AI (Gen-3 and successors) with API and studio tooling for media and creative applications. While not focused on physical world models, Runway overlaps with Odyssey's interactive video generation and serves similar creative and enterprise segments.

TypeBroad incumbent
Description

Meta's V-JEPA and related world-model research pursue similar goals of learning physical-world representations from video. Meta is a broad incumbent with significantly larger resources, competing with Odyssey in the underlying technology race even if its primary GTM is open research rather than a paid API.

Market position
Strengths5 records

Each record includes

Headline, Details, Source

Weaknesses5 records

Each record includes

Headline, Details, Source

Competitive moat5 records

Each record includes

Type, Details

Key risks6 records

Each record includes

Headline, Details, Source

Key highlights7 records

Each record includes

Headline, Details, Source

Customer concentration

Classification, Details

Segment5 records

Each record includes

Title, Type, Primary, Description, Pain point addressed, Use case, Source

Ideal customer profile6 records

Each record includes

Profile, Firmographic size, Sales motion, Sales cycle length, Buying structure, Purchase trigger, Buyer persona, Geography, Industry vertical, Primary use case, Description, Pain points, Evidence proof points, Target buyer

Technology focused
Yes
API detail
Has APIbool
Yes

Docs URL, Description

AI capability6 records

Each record includes

Type, Description, Source

AI maturity
App detail

Has app

Feature5 records

Each record includes

Title, Differentiator, Description, Source

Core technology
Revenue estimate
Valuation estimate
Number of profiles
Profiles3 records

Each record includes

Name, Designation, Designation category, Overview, Profile commentary, Source

No data
No data
Funding overview

Funding stage, Last funding date, Total funding USD

Funding rounds4 records

Each record includes

Round, Amount USD, Date, Pre money valuation, Total investors, Investors, News

Investors14 records

Each record includes

Name, Type, Date of entry, Rounds participated, Website

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Each record includes

Name, Acquisition type, Announced date, Completed date, Status, Website, News

Investment

Each record includes

Name, Round, Announced date, Lead investor, Website, News

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Odyssey

Generative AI / World Model Simulationodyssey.world

Odyssey is a private AI lab developing general-purpose world models—causal, multimodal systems that simulate physical environments with accurate physics—targeting robotics, gaming, healthcare, defense, and scientific research applications via API access and an AWS preferred-cloud partnership.

What Odyssey does

Odyssey is a private AI research lab founded in 2023 by Oliver Cameron and Jeff Hawke, both veterans of the autonomous vehicle industry. The company develops general-purpose world models—causal, multimodal AI systems trained primarily on video data to simulate physical environments with accurate physics. Its product portfolio comprises four named systems: Odyssey-2, a flagship world model that generates long-running interactive video simulations from text and image prompts with 50ms frame latency; Starchild-1, a model that learns from richer multimodal interaction beyond visual observation; Agora-1, a multi-agent world model enabling real-time human and AI co-inhabited simulations; and PROWL, a reinforcement-learning-driven adversarial research framework. The company is headquartered in Palo Alto with additional offices in Los Angeles, London, and Zurich, and employs approximately 55 people as of June 2026.

The underlying technology is differentiated by frontier claims in physical accuracy, long-horizon interactivity (multi-minute streaming versus fixed 10-second clips), and a documented cost advantage of achieving 80% model FLOP utilization on AWS Trainium3 chips—approximately double the 40-50% industry average. Odyssey targets five horizontal verticals: robotics (primary), gaming and interactive entertainment, healthcare and life sciences, defense and security, and scientific research, with use cases spanning robot task rehearsal, procedural game-world generation, medical training simulation, defense scenario generation, and physics-based scientific modeling.

The business model is API-first with an enterprise sales overlay. Odyssey offers API access to its world models through its website (odyssey.ml) and distributes through a strategic preferred-cloud partnership with Amazon Web Services, which also invested in the company's $310 million Series B in June 2026 (led by Natural Capital, with Amazon, AMD Ventures, Google Ventures, EQT, and In-Q-Tel participating) at a $1.45 billion post-money valuation. Revenue mechanics and pricing tiers are not publicly disclosed, and no customer logos have been disclosed in available sources. The company also benefits from sovereign AI infrastructure access through the UK government's AI Research Resource (Isambard-AI and Dawn supercomputers) as part of a £500M national AI initiative.

Odyssey firmographics

Firmographics
Name
Odyssey
Legal name
Odyssey
Website
https://odyssey.world
Company type
Private
Founded year
2023
Operating status
Operating
Headcount range
51–100 employees
Short description
Odyssey is a private AI lab developing general-purpose world models—causal, multimodal systems that simulate physical environments with accurate physics—targeting robotics, gaming, healthcare, defense, and scientific research applications via API access and an AWS preferred-cloud partnership.
Ownership category
akta.pro rank

Odyssey industry classification

Industry
Product category
Generative AI / World Model Simulation
NAICS
Software Publishers (513210)
SIC
Services-Prepackaged Software (7372)
akta.pro industry
PRIMARYHDAEANAA
End-to-End Enterprise AI Platforms (MLOps & Model Lifecycle Management)
Hardware, Technology & Telecom/Enterprise Platforms & Databases/Artificial Intelligence Platforms for Enterprise
SECONDARYHDAEANAC
Model Serving, Inference & Deployment Platforms (APIs, Edge/On-Prem)
Hardware, Technology & Telecom/Enterprise Platforms & Databases/Artificial Intelligence Platforms for Enterprise

Keywords

  • Generative world models
  • AI simulation platform
  • Physics-based video generation
  • Multimodal AI models
  • Developer API access

Where Odyssey is headquartered

Location

Headquarters

HQ city
Palo Alto
HQ country
United States
HQ region
North America

Offices3 records

Markets served

Odyssey business model

Business model
GTM type
B2B
Offering type
Software
Cost components
Technology or R&D, Infrastructure, Personnel, Marketing or Sales

Revenue model

  1. API Access / Model Licensing: Odyssey provides API access to its world model technology, suggesting a usage-based or subscription licensing model for developers and enterprises integrating world simulation capabilities.

Go-to-market motion2 records

Distribution channels2 records

Marketing channels3 records

Odyssey product offering

Product offering

Core offering

Odyssey is an AI lab that develops general-purpose world models—causal, multimodal AI systems that learn from video data to predict and simulate physical environments with accurate physics. Its flagship Odyssey-2 model generates long-running, interactive video simulations from text and image prompts, while complementary models (Starchild-1, Agora-1, PROWL) extend capabilities into multimodal learning, multi-agent simulation, and reinforcement-learning-driven adversarial training. The technology is delivered to developers and enterprises via API access and through a strategic AWS cloud partnership.

Product overview

Odyssey is an AI lab developing general-purpose world models—causal, multimodal systems that learn to predict and interact with the physical world over long horizons. The product portfolio centers on world model technology, with four key offerings: Odyssey-2 (flagship general-purpose world model generating interactive video simulations), Starchild-1 (multimodal world model learning from richer interaction), Agora-1 (multi-agent world model for real-time collaborative simulation), and PROWL (reinforcement learning research framework). These products form an integrated stack enabling applications in robotics, gaming, healthcare, and scientific simulation.

Differentiator

Problem solved

Functional benefit

Brands

  • Odyssey-2: Most powerful general purpose world model that materially advances state-of-the-art in physical accuracy of world models
  • Starchild-1
  • Agora-1
  • PROWL

Products and services

  • Odyssey-2

Quantifiable outcome

  • 80% model flop utilization on Trainium3 chips (vs 40-50% industry average)
  • +2 more outcomes

Companies that use Odyssey

Customer profile

Segments5 records

Ideal customer profiles6 records

Odyssey technology and API

Technology

Technology focussed Yes

API detail

Has API
Yes
API docs
API detail

Core technology

AI maturity

App detail

AI capability6 records

Feature5 records

Odyssey partnerships and signals

Strategic signal

Partnerships

One partnership is on record.

  • Amazon Web Services (AWS)coreTechnology or Integration · 17 June 2026Strategic partnership making AWS Odyssey's preferred cloud provider. Odyssey will use Amazon Trainium chips for compute-intensive world model training workloads. This is a key strategic relationship for both companies.

Scale indicators6 records

Recent moves6 records

Expansion highlights8 records

Odyssey competitors and assessment

Company assessment

Emerging players

  • Pika: Pika develops generative video models with an emphasis on controllable, interactive video creation. It shares Odyssey's modality (generative video) and a portion of its target use cases (creative content, simulation-style previews), though Pika is more focused on creative tooling than physical-world simulation.
  • Hailuo AI (MiniMax): MiniMax's Hailuo AI offers high-quality generative video models with a developer and consumer-facing API, competing in the same multimodal generative video category as Odyssey's world models. It is an emerging player with overlapping capabilities in text/image-to-video generation.
  • Lightricks: Lightricks offers AI-powered video generation tools (LTX Studio, etc.) targeting creators and marketers. It is a smaller, emerging player with partial overlap to Odyssey in generative video and interactive media, though focused more on creative workflows than physics-accurate world simulation.
  • Runway: Runway offers generative video AI (Gen-3 and successors) with API and studio tooling for media and creative applications. While not focused on physical world models, Runway overlaps with Odyssey's interactive video generation and serves similar creative and enterprise segments.

Direct peers

  • World Labs: World Labs is a direct competitor developing foundation world models for spatial intelligence and 3D scene generation, with the same core thesis of generative world models as Odyssey. Both target enterprise applications including robotics, VFX, and simulation with API-accessible models.
  • Decart: Decart builds real-time generative world models and interactive video environments, mirroring Odyssey's focus on long-horizon, interactive simulation. The companies share a target customer base of developers and enterprises integrating world-model APIs into robotics, gaming, and creative tools.

Broad incumbents

  • Stability AI: Stability AI is a foundation-model company offering generative models across images, video, and audio via APIs. It competes with Odyssey in the broader generative AI infrastructure space and overlaps in serving developer and enterprise customers with multimodal generation capabilities.
  • Google DeepMind (Genie): Google DeepMind develops Genie and other foundation world models as part of its broader AI research portfolio. While not focused exclusively on world models like Odyssey, DeepMind's scale, compute access, and multimodal research make it the most consequential incumbent competitor in the same technology category.
  • NVIDIA (Cosmos): NVIDIA's Cosmos platform provides world foundation models for physical AI and robotics simulation, overlapping directly with Odyssey's robotics and simulation use cases. As a broader AI infrastructure incumbent with deep compute partnerships, NVIDIA can both compete with and enable Odyssey's go-to-market.
  • Meta GenAI (V-JEPA): Meta's V-JEPA and related world-model research pursue similar goals of learning physical-world representations from video. Meta is a broad incumbent with significantly larger resources, competing with Odyssey in the underlying technology race even if its primary GTM is open research rather than a paid API.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights7 records

Customer concentration

Odyssey social profiles

Digital presence

Odyssey financial estimates

Financial estimate

Revenue estimate

Valuation estimate

Odyssey leadership team

Management profile

Number of profiles

Profiles3 records

Odyssey funding detail

Funding detail

Funding overview

Funding rounds4 records

Investors14 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

Odyssey M&A and investment

M&A and investment

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Odyssey

What does Odyssey do?

Odyssey is an AI lab that develops general-purpose world models—causal, multimodal AI systems that learn from video data to predict and simulate physical environments with accurate physics. Its flagship Odyssey-2 model generates long-running, interactive video simulations from text and image prompts, while complementary models (Starchild-1, Agora-1, PROWL) extend capabilities into multimodal learning, multi-agent simulation, and reinforcement-learning-driven adversarial training. The technology is delivered to developers and enterprises via API access and through a strategic AWS cloud partnership.

Is Odyssey a public or private company?

Odyssey is a private company. It is classified as venture growth investor backed and is currently operating.

When was Odyssey founded?

Odyssey was founded in 2023. It employs 51 to 100 people.

Where is Odyssey based?

Odyssey is headquartered in Palo Alto, United States, in the North America region.

How does Odyssey make money?

One revenue line is on record: API Access / Model Licensing.

Who are Odyssey's main competitors?

Emerging players on record are Pika, Hailuo AI (MiniMax), Lightricks and Runway. Direct peers are World Labs and Decart. Broad incumbents are Stability AI, Google DeepMind (Genie), NVIDIA (Cosmos) and Meta GenAI (V-JEPA).

Does Odyssey have an API?

Yes. Odyssey offers API access to its world model technology. The homepage features 'Get API Access' links, suggesting a developer-facing API is available for accessing Odyssey-2 and other world models. Specific details about API type (REST/GraphQL), authentication, rate limits, or documentation URL are not specified in the available source material.

What industry is Odyssey in?

Odyssey's product category is Generative AI / World Model Simulation. Its primary akta.pro industry code is HDAEANAA, End-to-End Enterprise AI Platforms (MLOps & Model Lifecycle Management), with a secondary code of HDAEANAC, Model Serving, Inference & Deployment Platforms (APIs, Edge/On-Prem). Its NAICS code is 513210 and its SIC code is 7372.

Unlock the full company data

50 free credits on sign-up, no credit card required.

Contact sales
Live signals
Robotics & Automation NewsOdyssey launches world model to control robots, humanoids, vehicles and dronesOdyssey announced Odyssey-3, its most powerful world model, designed to control robots, humanoids, vehicles, and drones. The model has been adopted by Flexion and demonstrated capabilities across six domains, including autonomous driving in India. It aims to provide a foundation for general-purpose physical intelligence.US About AmazonWhat AWS's chief AI and technology officer told employees about the future of AIAt AWS's Global Meeting, Matt Wood outlined AI's future, with Odyssey demonstrating Odyssey-3 and Agora-2. Odyssey-3 runs on Trainium3 at over 80% model flops utilization, and Agora-2 supports 20 participants in a shared simulation. Wood said AWS aims to make building AI systems as easy as building websites.Tech TimesOne Backbone, Five Bodies: Odyssey-3 Drives Cars, Runs Humanoids, Plays GamesOdyssey announced Odyssey-3, a frozen backbone model that controls robot arms, humanoids, cars, drones, and game characters via lightweight decoders. A driving policy trained only on simulation retained 77% of real-world performance on Indian roads, and a GTA V policy controlled a horse in Red Dead Redemption 2. Independent benchmarks have not yet been published.StartupHub.aiWorld Models: The Next Frontier Beyond LLMsOdyssey, co-founded by Oliver Cameron, is positioning itself at the forefront of world model development to capture a significant market opportunity in foundational AI models. The company highlights that investors have poured over $3 billion into world model startups in the first half of 2026, driven by the need for systems that can understand physics and dynamics beyond text generation. Odyssey aims to apply these models across diverse industries including robotics, autonomous driving, and defense.OdysseyOdyssey Announces Investment from NVentures and Samsung NextOdyssey announced an investment from NVentures, NVIDIA's venture capital arm, and Samsung Next on February 12, 2026, to accelerate research towards a general-purpose world simulator. The funding will support development of their Odyssey-2 Pro model and scale the infrastructure needed to train and serve next-generation world models. The company, founded two years ago with researchers recruited from leading AI firms, aims to achieve what they describe as the 'GPT-3 of world models.'OdysseyIntroducing PROWL-1: Learning Through DiscoveryOdyssey has introduced PROWL-1 (Prioritized Regret-Driven Optimization for World Model Learning), a novel RL-driven adversarial framework designed to improve world models by having an RL agent discover failures in geometry, motion, visual consistency, and action-conditioned dynamics within game environments. The framework uses a Prioritized Adversarial Trajectory (PAT) buffer to create a progressively harder curriculum from discovered failures, converting them into targeted training trajectories that fine-tune the world model over time. The team evaluated PROWL-1 in the MineRL research environment using Minecraft, demonstrating improvements in action-following, visual quality, crosshair persistence, emergent behavior rendering, temporal coherence, and out-of-distribution robustness compared to baseline models.OdysseyAgora-1: The Multi-Agent World ModelOdyssey released Agora-1, the first multi-agent world model that enables multiple participants to share and interact within the same generated world simulation in real-time. The system uses a decoupled architecture that separates simulation dynamics from rendering, allowing consistent multi-view generation from a shared world state with applications across gaming, robotics, defense, and education. Agora-1 is an early research preview built using GoldenEye as a test environment, supporting up to four players in a shared deathmatch simulation.OdysseyStarchild-1: The First Real-Time Multimodal World ModelOdyssey announced Starchild-1, described as the world's first multimodal world model capable of autoregressively generating synchronized audio and video in real-time while continuously responding to streaming user input. The model addresses challenges of different temporal frequencies between audio and video through a new causal multimodal training and inference stack. The company released an accompanying technical report detailing the architecture, causal distillation pipeline, and asynchronous KV-cache architecture designed for long-horizon multimodal stability.OdysseyIntroducing Odyssey-1: A Playable World ModelOdyssey, an AI lab pioneering general-purpose world models, has launched Odyssey-1, a research preview of a real-time playable world model that generates interactive video at 30 FPS with 40ms latency, responding to user inputs. The model learns both pixels and actions from real-life video, distinguishing it from traditional video models that generate fixed clips. Infrastructure costs for the experience are currently $1-2 per user-hour, with the company positioning this as an early step toward real-time world simulation technology.OdysseyIntroducing Odyssey-2 Max: Scaled World SimulationOdyssey announced the launch of Odyssey-2 Max, its largest and most powerful general world model, which advances the state-of-the-art in physical accuracy of world simulations through next-state prediction at scale.