Sarbam processes 450 billion tokens per day... DeepSeek API released
Original headline사르밤, 하루 4500억 토큰 처리…딥시크 API 공개
Not investment advice. Muhabbit scores how much a story may matter, not which way to trade.
India's Sarvam launched Token Factory, a paid API platform that routes requests across its in-house Sarvam 105B model and third-party models including DeepSeek V4.1 Flash 1017, DeepSeek V4 Pro, Moonshot AI's Kimi K3.1 Pro, Z.AI's GLM 5.3, and Google's Gemma 4, with compute handled by GPU clusters located in India. The company said developers and enterprises can access both proprietary and external models through a single API, and DeepSeek V4.1 Flash 1017 was formally listed for general availability in Sarvam's API changelog on October 30. The platform is positioned to process up to 450 billion tokens per day, according to the report. The move matters because it makes Sarvam a multi-model aggregator competing with global inference providers while keeping workloads on domestic infrastructure. What to watch next is adoption of the paid tier, pricing and latency benchmarks across the hosted models, and whether Sarvam expands its model roster or GPU capacity.
Muhabbit's AI wrote this summary in its own words from TokenPost's report and translated it from Korean. It can get details wrong, so check the source.
Named in this story
Sarvam, DeepSeek, Token Factory, Google (GOOGL), Moonshot AI, Z.AI, India
A linked name opens that asset's sentiment page.