NVDA 145.98 ▼3.30%GOOGL 272.66 ▼3.46%MSFT 498.30 ▼1.78%AMD 489.19 ▼7.72%INTC 55.72 ▲1.92%TSMC 352.09 ▼2.90%AMZN 257.37 ▼3.87%META 596.22 ▼3.27%AAPL 349.21 ▼0.92%PLTR 153.68 ▼1.91%
Markets at last close

Models

SquadStack.ai launches Maya, claims lead in sales benchmark

·1 min read

SquadStack.ai has launched Maya, a language model specialized for sales conversations, claiming it outperformed models from OpenAI, Google, DeepSeek, Alibaba and NVIDIA in internal testing. Maya scored 96.7% on the company’s SquadStack Telesales LLM Benchmark, compared with 92.5% for Google’s Gemma 4 31B, 92.2% for DeepSeek V4.1 Flash and 89% for OpenAI’s GPT-6 Luna.

The company said it post-trained an open-weight foundation model on more than 40 crore real sales interactions collected over six years, with calls labeled by outcomes such as sales, callbacks and drop-offs. Maya is designed to handle informal responses, hesitation and interruptions in Indian sales conversations. SquadStack.ai says it currently handles more than 50 lakh conversations daily across Hinglish and Indian languages.

The in-house benchmark used 9,962 conversational moments from 471 real sales calls to assess models’ next responses. Anthropic’s Claude Haiku 5.5 judged responses against seven parameters, including maintaining conversational objectives and using tools correctly. SquadStack.ai said Maya maintained context on 96% of turns and selected an appropriate response 95% of the time.

Maya is built and hosted in India, which the company says could help banks, lenders and insurers address requirements under India’s Digital Personal Data Protection Act. Planned improvements include learning from pauses, changes in tone and customers’ previous interactions.

Originally reported by economictimes.indiatimes.comRead the source →
Related coverage