E2E Networks launches Blaze, pay-per-token inference in India
One OpenAI-compatible API for open-weight models, hosted in India and billed in INR. No public prices yet.
E2E Networks, an Indian cloud provider listed on the NSE and BSE, launched Blaze on November 2, 2026. Blaze is a pay-per-token inference service for open-weight models, served from E2E's NVIDIA GPU infrastructure in India through one OpenAI-compatible API. E2E disclosed the launch to India's stock exchanges the same day.
What changed
- Blaze has one OpenAI-compatible endpoint,
https://blaze.e2enetworks.net/v1, for open-weight text, voice, OCR, multimodal and diffusion models. - The press release names NVIDIA Nemotron, Sarvam, DeepSeek, Qwen, GLM and Gemma. The product page lists models including Qwen3.8 Flash, Qwen3.8 Max, DeepSeek V4 Flash, Kimi K3, GLM 5.3, IBM Granite 4.2 8B, Tencent's Hy-MT2 translation models, and Sarvam 30B and 100B.
- It runs in Indian data centres and is billed in INR. Customers pay for the tokens they use, with no GPU capacity to reserve.
- Sending the header
x-model: autowith"model": "auto"lets Blaze's routing layer choose the model for each request. - The filing lists Blaze for domestic and international customers. E2E has not published per-token prices, and the product page asks new users to request access.
Sample catalogue
From E2E's product page, in tokens:
| Model | API alias | Context | Max output |
|---|---|---|---|
| Qwen3.8 Flash | qwen/qwen3.8-flash | 1,000,000 | 131,072 |
| GLM 5.3 Flash | z-ai/glm-5.3-flash | 1,310,720 | 131,072 |
| DeepSeek V4 Flash Vision Exp | deepseek/deepseek-v4-flash-vision-exp | 1,048,576 | 384,000 |
| Granite 4.2 8B | ibm-granite/granite-4.2-8b | 131,072 | 117,964 |
| Hy-MT2-1.8B | tencent/hy-mt2-1.8b | 8,192 | 4,096 |
What it means
Teams that need inference to stay in India get a pay-per-token option for widely used open-weight models without leasing GPUs. Until E2E publishes per-token prices, its cost cannot be compared with other hosts.
The same models on RouterPlus
Blaze is not a provider in the RouterPlus catalog. Lowest input and lowest output prices for the same models in the catalog, November 3, 2026, per million tokens:
| Model | Lowest input | Lowest output |
|---|---|---|
| Qwen3.8 Flash | $0.113 | $0.382 |
| Qwen3.8 Max | $1.65 | $4.951 |
| GLM 5.3 Flash | $0.05 | $0.20 |
| GLM 5.3 | $0.30 | $2.50 |
| DeepSeek V4 Flash | $0.14 | $0.28 |
| DeepSeek V4 Flash Vision Exp | $0.2156 | $0.6468 |
| Kimi K3 | $1.50 | $9.50 |
| Granite 4.2 8B | $0.06 | $0.25 |
Compare every model and provider on the RouterPlus models page.
What to do
- If you need inference hosted in India, request access and ask E2E for per-token prices and the models live on your account.
- Pin a model ID in production if you need stable outputs.
autolets Blaze pick the model per request.