# Sarvam AI — Introducing Sarvam Prism

- Company: Sarvam AI (sarvam.ai)
- Announced: 2026-11-03
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://www.sarvam.ai/blogs/sarvam-prism/
- Record: https://forck.live/items/31005-introducing-sarvam-prism
- Subject: Sarvam / Bulbul / Saaras
- Open weights: yes
- Models affected: Sarvam Prism, Sarvam 105B, Sarvam 105B-10B, Jev 1.14, Perplexity Decider v1.2, Clef, Microsoft-Decision-1, GLiDE 1.2, mammoth-1

Sarvam is releasing Sarvam Prism, a System One decision model built on Sarvam 105B, aimed at enterprise workflows that are sequences of fast decisions. Prism scores 71.8 while responding in 44 ms on a B200 GPU, which Sarvam says is the highest score under 100 ms across the 25 decision models it tested. The company says Prism makes fewer errors than its confidence scores predict across thresholds from 50% to 95% in English and Indian languages, and that its answers stay the same when options are shuffled. Sarvam is releasing Prism with open weights so enterprises can train the calibrated base on their own data and run it on their own infrastructure. On English answers at 90% confidence or higher, Sarvam says Prism predicts 3.3% errors and makes 2.2%, while Perplexity Decider, Microsoft-Decision-1 and Jev 1.14 underestimate their errors by factors of 1.5, 1.8 and 2.7. Sarvam says it tested Prism on use cases including speech to speech harness, trace evaluations, prompt injection guardrails, intent routing for customer support, model routing and document intelligence.

## Evidence

Verbatim from https://www.sarvam.ai/blogs/sarvam-prism/:

> Today we are releasing Sarvam Prism, a System One decision model built on Sarvam 105B. Typeface created this category last month and named their Jev model after Kahneman’s System One thinking, the fast and intuitive mode of judgement that handles most of what people decide in a day. Much of work in an enterprise is a sequence of such decisions, and well calibrated decision models fit neatly into these workflows.
> 
> Best-in-class accuracy & speed: Prism sets a new point on the Pareto frontier of accuracy and speed among the 25 decision models we tested. The model scores 71.8 while responding in just 44 ms on B200 GPU, the highest-scoring model under 100 ms across the 25 models we tested, averaged over three indices. We believe System One models improve with scale and benefit from mixture-of-expert architecture trade-offs, hence Prism is trained on the Sarvam 105B-10B base.
> 
> Calibration enterprises can trust: For an enterprise to automate a workflow, it has to trust the model’s confidence on its output. Prism’s errors is always within its confidence bounds and its answer is not affected by the order in which options are presented.

---

Record: https://forck.live/items/31005-introducing-sarvam-prism
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md