iCreatAI News
Skip to content
MarkTechPost· Asif Razzaq·· 16 hr agoAI score60

Sarvam Launches Sarvam Token Factory: Bringing India-Hosted Inference to 8 Open Models

Sarvam Launches Sarvam Token Factory: Bringing India-Hosted Inference to 8 Open Models

AI summary

Sarvam has opened Sarvam Token Factory, an inference platform serving its own models and open-weight models from NVIDIA B200 GPU clusters in India through a single OpenAI-compatible API, with 8 models listed at launch including Sarvam 105B, GLM 5.3, DeepSeek V4.1 Flash 1017, Kimi K3.1 Pro and Gemma 4 31B. Sarvam reports that Sarvam 105B fits on one B200 per replica and that Kimi K3.1 Pro reaches 6,177 tokens/s per GPU versus 3,213 for InferenceX at 70 tokens/s per user, though Together posts a faster first token (0.71 s vs 0.90 s) and higher streaming speed (278 vs 130 tokens/s) on DeepSeek V4.1 Flash with a 10K prompt. Introductory pricing runs through the end of 2026 at ₹10/₹25 per 1M tokens for Sarvam 105B and ₹14/₹55 for DeepSeek V4.1 Flash 1017, and all benchmark numbers are self-reported by Sarvam.

Source: MarkTechPost · marktechpost.com