Announcement

On Friday, 16 October 2026, Sthanika – the applied‑AI research lab of the PurpleTalk group based in Hyderabad, Telangana – announced that its Sieve 31B decision model secured the second position globally on the Jev Decision Index, an open benchmark hosted on Hugging Face for rapid‑decision AI models.

Performance Details

  • Full score: 63.8 (combined public test suite and private tests conducted by the benchmark maintainers).
  • Public‑test score: 62.5, placing second on the public portion of the leaderboard.
  • The model outperformed both Jev and Perplexity Decider v1.1 in the ranking.
  • Private tests, which draw from domains outside the public suite, were highlighted as the most significant indicator of the model’s generalisation capability.

Model Characteristics and Cost Efficiency

Sieve 31B is described as a cost‑efficient decision model that can be invoked as a single component within larger AI systems. Unlike general‑purpose large language models (LLMs) that are optimized for conversational tasks and are relatively expensive when used for repetitive, defined‑option decisions, Sieve 31B delivers decisions more reliably, faster, and at a fraction of the cost. The model reads a question and its presented choices, returns a definitive answer together with a confidence score, and enables software to audit each decision – attributes critical for applications where consistency, speed, and traceability matter.

Sieve Family Rankings

The broader Sieve family – comprising Sieve 2B Plus, Sieve 4B, and Sieve 9B Plus – also achieved top‑rank positions within their respective size classes on the same Jev Decision Index leaderboard.

Current Deployments and Future Vision

Sieve models are already integrated into PurpleTalk products Ello and Mina, where they handle routine judgments more economically than generative LLMs. The lab envisions a new category of agentic systems that are more reliable, controllable, and low‑latency, targeting use‑cases such as voice interfaces, customer‑experience automation, and robotics.

Executive Quotes

  • Sridhar Muppidi, Co‑founder of PurpleTalk, stated: “Sieve 31B shows that a focused model fine‑tuned in India can perform among the best in the world at this work. The private tests matter most to us. They come from domains outside the public suite, and Sieve generalised better than most of the field. It learned the task, not the test.”
  • Bharat Lingam, CEO & Co‑founder of PurpleTalk, added: “We believe decision models will open up a new category of agentic systems: more reliable, more controllable and fast enough for voice, customer experience and robotics. We are publishing everything needed to reproduce our results, and we invite the research community to test it, challenge it and build on it.”

Availability

All models and evaluation artifacts are publicly available via the provided links, and full benchmark results can be accessed on the Jev Decision Index website.

About Sthanika and PurpleTalk

Sthanika, whose name means “local” in Sanskrit, is a Hyderabad‑based AI research lab that builds open‑weight models for tasks that general‑purpose models were never designed to solve. It operates under the PurpleTalk group, founded in 2008, which develops AI companies and services across sectors such as healthcare, agriculture, financial services, and retail. PurpleTalk’s portfolio includes (x)cube LABS, Sthanika, Ello, and Mina.

---