Turing
Turing Inc. is a Tokyo-based AI company founded in 2021 developing camera-only End-to-End autonomous driving systems using Vision-Language-Action foundation models, targeting Japanese automotive OEMs and government-backed research programs.
- Company typePrivate
- Founded2021
- HeadquartersHigashi-shinagawa, Japan
- Headcount51–100
- GTM typeB2B
- OfferingSoftware
What Turing does
Turing Inc. (チューリング株式会社) is a Tokyo-headquartered AI company founded in August 2021 that is developing Level 5 (fully autonomous) driving technology. The company serves Japan's automotive sector and operates a pre-commercial model funded by government research grants and equity capital. In January 2026, Turing relocated its headquarters to Heiwajima, Ota-ku, Tokyo.
The company's core product is a camera-only autonomous driving stack built on an End-to-End (E2E) AI architecture fused with Vision-Language-Action (VLA) foundation models. Perception, path planning, and vehicle control are unified within a single neural network rather than separated into modular pipelines. Proprietary technology assets include approximately 2-billion-parameter VLA models running at 10Hz real-time inference, the RACER causal reasoning dataset, the DriveTiTok image tokenizer, the Terra generative world model, the Heron multimodal learning library, the CoVLA Dataset, the Japan Open Driving Dataset (~111 hours, 24TB), and Llama-3-heron-brain Japanese LLMs (70B/8B).
Turing is pre-commercial and generates no product revenue. Income derives from Japanese government grants — primarily through the METI/NEDO GENIAC program, where Turing has been selected for all four phases — alongside an estimated ¥7.2 billion in debt and approximately ¥20 billion in equity funding (cumulative ~¥27.2 billion). Go-to-market is structured around technology licensing and joint development with Japanese automotive OEMs; current partners include Subaru and DENSO under joint research agreements on E2E and physical foundation models.
Turing firmographics
Firmographics- Name
- Turing
- Legal name
- Turing Inc.
- Website
- https://tur.ing
- Company type
- Private
- Founded year
- 2021
- Operating status
- Operating
- Headcount range
- 51–100 employees
- Short description
- Turing Inc. is a Tokyo-based AI company founded in 2021 developing camera-only End-to-End autonomous driving systems using Vision-Language-Action foundation models, targeting Japanese automotive OEMs and government-backed research programs.
- Ownership category
- akta.pro rank
Turing industry classification
Industry- Product category
- Autonomous Driving AI Software
- NAICS
- Search, Detection, Navigation, Guidance, Aeronautical, and Nautical System and Instrument Manufacturing (334511), Computer Systems Design and Related Services (54151), Computer Systems Design and Related Services (5415)
- SIC
- Search, Detection, Navagation, Guidance, Aeronautical Sys (3812), Services-Computer Programming Services (7371)
- akta.pro primary industry
- Autonomous Driving & ADAS Perception (HDAAAEAI)
- akta.pro secondary industries
- Open-Source Model Ecosystems & Model Marketplaces (HDAAACAM), Real-Time Vision at the Edge (on-device CV/video) (HDAAAJAE), Model Compression & Optimization (Quantization, Distillation, Pruning) (HDAAACAL), Fine-Tuning, Adaptation & Custom Model Training (PEFT/LoRA/RLHF) (HDAAACAC)
Keywords
Where Turing is headquartered
LocationHeadquarters
- HQ city
- Higashi-shinagawa
- HQ country
- Japan
- HQ region
- Asia
Offices1 record
Markets served
Turing business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Technology or R&D, Personnel, Infrastructure, Marketing or Sales, Operations
Revenue model
- Government Grants and Subsidies: Turing receives government funding through METI/NEDO programs (GENIAC) and regional autonomous driving service development subsidies. These grants support R&D for foundation models, computing resources, and dataset construction.
- Autonomous Driving System Licensing / Technology Licensing (Future): Turing is developing fully autonomous driving technology for eventual licensing to automotive manufacturers. Joint research with Subaru and DENSO represents early commercial engagement toward this revenue stream.
- Autonomous Driving-as-a-Service / Subscription (Future): Based on industry trends (Tesla's autonomous driving subscription model), Turing anticipates recurring subscription revenue from autonomous driving capabilities deployed in vehicles at scale.
Go-to-market motion1 record
Distribution channels3 records
Marketing channels10 records
Turing product offering
Product offeringCore offering
Turing develops a camera-only End-to-End (E2E) autonomous driving AI system that fuses environmental perception, path planning, and vehicle control into a single neural network, layered with a Vision-Language-Action (VLA) foundation model for contextual and common-sense reasoning. The company builds supporting AI infrastructure including the Heron multimodal learning library, the Terra generative world model, the DriveTiTok image tokenizer, Japanese LLMs (Llama-3-heron-brain 70B/8B), and large-scale proprietary Japanese driving datasets (CoVLA, Japan Open Driving Dataset, RACER) for VLA training. The technology is delivered via government-funded R&D programs, joint research with automotive OEMs (Subaru, DENSO), and open-source releases to the research community, with future plans for licensing and subscription commercialization.
Product overview
Turing develops a unified E2E (End-to-End) autonomous driving system as its core product, using camera-only input to make all driving decisions without LiDAR or HD maps. The portfolio includes the Heron multimodal learning library for vision-language integration, Terra generative world model for realistic driving simulation, CoVLA and Japan Open Driving datasets for model training, DriveTiTok image tokenizer for efficient visual compression, and Japanese LLMs (Llama-3-heron-brain series). VLA models integrate these components for real-time autonomous driving with language understanding capabilities.
Differentiator
Problem solved
Functional benefit
Products and services
- E2E (End-to-End) Autonomous Driving System Camera-only autonomous driving AI that makes all driving decisions (steering, acceleration, braking) through a single neural network, without relying on LiDAR or HD maps. Integrates environmental perception, path planning, and vehicle control into a unified model. Used by automotive OEMs (Subaru, DENSO) for joint research and by METI/NEDO-funded programs.
- Heron Multimodal Learning Library Multilingual (Japanese/English) large-scale multimodal learning library that connects image recognition models with large-scale language models. Supports dialogue-based dataset training for natural conversational responses and is foundational to Turing's VLA architecture. Open-sourced under Apache License 2.0.
- Terra — Generative World Model for Autonomous Driving Japan's first generative world model for autonomous driving that understands complex real-world physics and object interactions, producing realistic driving scenes as video. Can be used as a driving simulator and as a component of autonomous driving systems for traffic analysis and prediction. Trained on ~1,500 hours of proprietary driving data.
- CoVLA Dataset Comprehensive Vision-Language-Action Dataset for Autonomous Driving — Japan's first VLA model dataset, comprising over 80 hours of driving data with diverse annotations. Enables detailed natural language descriptions of driving environments and route planning. Released on Hugging Face for research use.
- Japan Open Driving Dataset Approximately 111 hours (~24TB) of Japanese driving data in nuScenes-compatible format. Includes caption data for VLM/VLA training covering perception, prediction, and planning. Released open-source for E2E autonomous driving and VLA research.
- DriveTiTok Image Tokenizer Image tokenizer that converts driving video into discrete tokens, compressing data to approximately 1/100th of original size while preserving critical visual information for driving decisions through temporal dynamics and scene context. Released on Hugging Face.
- RACER Dataset (Rationale-Aware Captioning of Edge-Case Driving Scenarios) Causal reasoning dataset for VLA models that captures the rationale behind driving decisions through structured causal descriptions, enabling AI to understand why specific actions are taken and generate more appropriate and reliable behavior in edge cases. Released on Hugging Face.
- Llama-3-heron-brain Japanese LLM (70B / 8B) Instruction-tuned Japanese LLM based on Llama-3, optimized for autonomous driving multimodal models. Available in 70B and 8B parameter versions. Outperformed GPT-3.5-turbo on the shared GENIAC benchmark and achieved the second-highest score among participants in Japanese/English language performance. Released on Hugging Face.
- vlm-recipes Distributed Learning Library High-speed multimodal distributed learning library for training large-scale vision-language multimodal models using the PyTorch FullyShardedDataParallel (FSDP) backend, supporting distributed training across large GPU clusters. Released open-source on GitHub.
Quantifiable outcome
- First in Japan: VLA model real-time autonomous driving on public roads at 10Hz
- +4 more outcomes
Companies that use Turing
Customer profileNamed customers3 records
Segments3 records
Ideal customer profiles3 records
Turing technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- No
- API docs
- API detail
Core technology
AI maturity
App detail
AI capability10 records
Feature9 records
Turing partnerships and signals
Strategic signalPartnerships
Five partnerships are on record, tiered core.
- Subaru CorporationcoreJoint research on in-vehicle End-to-End (E2E) autonomous driving system implementation. Turing and Subaru collaborate on integrating a lightweight E2E autonomous driving model with vehicle control systems, with a view toward future public-road demonstration. This partnership is conducted under the GENIAC program (METI/NEDO).
- DENSO CorporationcoreJoint research on the development of a physical foundation model for autonomous driving. Based on vision-language models and vision state-space models, the companies advance technology development to acquire linguistic and spatio-temporal understanding from camera sensors alone, with adaptation to actual vehicles. This partnership is conducted under the GENIAC program (METI/NEDO).
- HumaincoreStrategic partnership to build the world's first enterprise AI Agent Marketplace on HUMAIN ONE, announced during the FII Priority Summit in Miami. The marketplace enables organizations to discover, deploy, and scale AI agents across business functions. Supports Saudi Arabia's goal to become a global AI hub. Note: This appears to involve Turing Inc. (US-based AI firm), which may be a separate entity from Turing Corporation (Japan-based autonomous driving company).
- METI / NEDOcoreSelected as joint applicant for FY2024 supplementary subsidy program for development and demonstration of autonomous driving services addressing regional mobility challenges. Part of 'Regional Mobility DX Strategy'.
- METI / NEDOcoreMinistry of Economy, Trade and Industry and New Energy and Industrial Technology Development Organization. Selected as grant recipient for the GENIAC program to develop competitive generative AI foundation models for fully autonomous driving. Received ¥740 million in GPU computing resources.
Scale indicators16 records
Recent moves7 records
Expansion highlights6 records
Turing competitors and assessment
Company assessmentMarket position
Competitive moat5 records
Key risks6 records
Key highlights7 records
Customer concentration
Turing social profiles
Digital presenceTuring financial estimates
Financial estimateRevenue estimate
Valuation estimate
Turing leadership team
Management profileNumber of profiles
Profiles10 records
Turing funding detail
Funding detailFunding overview
Funding rounds11 records
Investors58 records
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Turing M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Turing
What does Turing do?
Turing develops a camera-only End-to-End (E2E) autonomous driving AI system that fuses environmental perception, path planning, and vehicle control into a single neural network, layered with a Vision-Language-Action (VLA) foundation model for contextual and common-sense reasoning. The company builds supporting AI infrastructure including the Heron multimodal learning library, the Terra generative world model, the DriveTiTok image tokenizer, Japanese LLMs (Llama-3-heron-brain 70B/8B), and large-scale proprietary Japanese driving datasets (CoVLA, Japan Open Driving Dataset, RACER) for VLA training. The technology is delivered via government-funded R&D programs, joint research with automotive OEMs (Subaru, DENSO), and open-source releases to the research community, with future plans for licensing and subscription commercialization.
Is Turing a public or private company?
Turing is a private company. It is classified as venture growth investor backed and is currently operating.
When was Turing founded?
Turing was founded in 2021. It employs 51 to 100 people.
Where is Turing based?
Turing is headquartered in Higashi-shinagawa, Japan, in the Asia region.
How does Turing make money?
Three revenue lines are on record. Government Grants and Subsidies are the primary driver. The others are autonomous Driving System Licensing / Technology Licensing (Future) and autonomous Driving-as-a-Service / Subscription (Future).
Does Turing have an API?
No public API is recorded for Turing.
What industry is Turing in?
Turing's product category is Autonomous Driving AI Software. Its primary akta.pro industry code is HDAAAEAI, Autonomous Driving & ADAS Perception, with a secondary code of HDAAACAM, Open-Source Model Ecosystems & Model Marketplaces. Its NAICS code is 334511 and its SIC code is 3812.