Fordel StudiosIndian AI lab Sarvam’s new models are a major bet on the viability of open source AI
What Happened
The new lineup includes 30-billion- and 105-billion-parameter models; a text-to-speech model; a speech-to-text model; and a vision model to parse documents.
Our Take
Look, open source AI from an Indian lab landing 30B and 105B parameter models is legitimately undermining the "only US can build this" narrative. That matters for geopolitics.
Technically? They're hitting the same spot as Llama — usable but not frontier. The real win is timing: they're releasing when enterprise adoption is desperate for non-OpenAI options.
But here's the thing — open source models are commoditizing. If Meta can do 405B parameters on open source, can Sarvam compete long-term?
What To Do
Evaluate Sarvam models for Indian regulatory compliance use cases only — that's their actual moat.
Cited By
React