News ArticleE2E Networks Introduces Blaze, a Pay-Per-Token AI Inference Service with NVIDIA GPU Power
Launching November 2, 2026, Blaze offers India-hosted, OpenAI-compatible API access to leading open AI models worldwide
E2E Networks launches Blaze, a global pay-per-token AI inference service for open AI models, delivering NVIDIA GPU-powered, India-hosted AI infrastructure for enterprises and developers. Blaze supports multiple open AI models and offers scalable, latency-optimized inference across domestic and international markets.
E2E Networks Limited
New Ventures
E2E Networks Limited has launched Blaze, a new pay-per-token AI inference service providing access to open AI models globally via an OpenAI-compatible API powered by NVIDIA GPU infrastructure hosted in India.
Blaze is designed for enterprises and developers requiring efficient, scalable AI inference for workloads such as coding agents, voice applications, document AI, and image generation. It emphasizes low latency, token cost efficiency, and supports long-context workloads through custom NVIDIA CUDA kernels and Dynamo-based serving.
The product supports open models including NVIDIA Nemotron, Sarvam, DeepSeek, Qwen, GLM, and Gemma. Blaze is built on the NVIDIA DSX OS platform, incorporating an architecture that ensures hardware-level isolation for sensitive sectors like banking and healthcare, complying with strict Indian data sovereignty requirements.
Blaze allows customers to pay only for tokens used and integrates easily with other E2E Cloud platform services. It complements E2E’s AI platform TIR, which provides Kubernetes-managed infrastructure for AI training and inference, running on India-based NVIDIA Blackwell GPU clusters.
E2E Networks is expanding its AI factory capacity to about 14.6 MW, featuring advanced Blackwell GPU clusters to support AI workload scaling, reinforcing India-hosted sovereign compute capabilities.
Highlights
- Blaze launched on November 2, 2026, as a pay-per-token inference service for open AI models.
- It provides a single OpenAI-compatible API endpoint for multiple open models globally.
- Powered by NVIDIA GPU infrastructure hosted in India, meeting data sovereignty needs.
- Designed for AI workloads demanding low latency and efficient token use, including coding and voice applications.
- Integrates NVIDIA DSX OS components for scalable, secure AI serving with hardware isolation.
- Blaze complements E2E’s TIR platform for training and fine-tuning with 14.6 MW of AI factory capacity planned.
Key Data Snapshot
What happened
- E2E Networks launched Blaze, a pay-per-token AI inference service, on November 2, 2026.
- Blaze provides an OpenAI-compatible API endpoint for multiple leading open AI models globally.
- The service is powered by NVIDIA GPU infrastructure hosted in India, supporting sovereign AI compute needs.
Key details
- Blaze supports domestic and international markets with global availability.
- It targets use cases such as coding agents, long-context AI workloads, voice and chat applications, document AI, and image/video generation.
- Service features custom NVIDIA CUDA kernels and NVIDIA Dynamo-based serving for performance optimization.
- Hardware-level isolation with NVIDIA BlueField DPUs and DOCA ensures security compliance for sensitive industries.
Operational performance
- Blaze offers fast latency to first generated sentence suitable for latency-sensitive applications.
- High key-value cache reuse reduces inference costs for open-weight models.
- Customers pay only for tokens consumed, enhancing cost efficiency.
Strategic context
- Blaze utilizes NVIDIA DSX OS open-source components as part of a modular AI Factory Platform.
- E2E Networks also operates TIR, a Kubernetes-managed platform for AI model training and self-hosted inference.
- E2E's AI factory capacity includes advanced Blackwell GPU clusters totaling about 14.6 MW.
- Blaze extends E2E’s vision of providing sovereign AI infrastructure for Indian AI developers and enterprises.
Background and context
- The launch addresses increasing token consumption and cost sensitivity as AI moves to production use cases in India.
- Maintaining data sovereignty and compute infrastructure within India is a priority for customers across sectors.
- E2E Networks operates as an NVIDIA Cloud Partner, leveraging NVIDIA reference design clusters for sovereign computing.
Highlights
- Blaze launched on November 2, 2026, as a pay-per-token inference service for open AI models.
- It provides a single OpenAI-compatible API endpoint for multiple open models globally.
- Powered by NVIDIA GPU infrastructure hosted in India, meeting data sovereignty needs.
- Designed for AI workloads demanding low latency and efficient token use, including coding and voice applications.
- Integrates NVIDIA DSX OS components for scalable, secure AI serving with hardware isolation.
- Blaze complements E2E’s TIR platform for training and fine-tuning with 14.6 MW of AI factory capacity planned.
Disclosures
- SEBI (Listing Obligations and Disclosure Requirements) Regulations, 2015
- SEBI Master Circular No. HO/49/14/14 (7) 2025-CFD-POD2/I/3762/2026 dated January 30, 2026
More News on New Ventures
View allLatest corporate announcements and market updates in New Ventures.
Coforge Launches BridgeCX to Deliver AI-Powered Real-Time Personalized Customer Experiences
Coforge Limited has launched BridgeCX, a next-generation customer engagement platform that leverages Pega Customer Decision Hub combined with Mastercard Dynamic Yield to provide real-time, hyper-personalized digital experiences for enterprises.
Read Full Story
Allied Blenders and Distillers' ABD Maestro unveils Rangeela Jhatpat Jamun and Kala Khatta super-premium vodkas inspired by Indian street flavors
ABD Maestro, the super-premium spirits subsidiary of Allied Blenders and Distillers Limited, has expanded its Rangeela Vodka lineup with two new Indian-flavored variants—Rangeela Jhatpat Jamun Vodka and Rangeela Kala Khatta Vodka—combining local street tastes with craft vodka production.
Read Full Story
Belrise Industries Secures License and Technical Assistance from Yamato Co. for Manufacturing Automotive Filters and Intake Manifolds in India
Belrise Industries Limited has entered into a License and Technical Assistance Agreement with Japanese company Yamato Co., Ltd to manufacture and sell automotive filtration products and resin intake manifolds in India under license and technical support.
Read Full Story
