India’s Frontier AI Race Gets a 240M Model That Beats Liquid AI
NeuroDX, backed by the IndiaAI Mission, launched Sphota, a language model based on a novel architecture focused on capturing meaning rather than next-token prediction. According to the company's preliminary benchmarks, Sphota scored 52.4% versus Liquid AI's 52.1% across ~11 tests, while using significantly less training data and achieving 4.5x better memory efficiency. The model grew out of the startup's work interpreting EEG signals for healthcare applications. Results remain unverified by independent evaluation.
Key points
- Sphota outscored Liquid AI (52.4% vs 52.1%) on ~11 benchmarks
- Requires 50x less training data and 4.5x better memory efficiency
- Novel architecture prioritizes meaning-capture over next-token prediction
Summary and key points written by AI from the headline and a short excerpt, so they can be incomplete or wrong. Analytics India Magazine did the reporting; read the original for the full story.
Read the full story at Analytics India Magazine
