Fast enough for real-time work, thoughtful enough for real problems. Vaani is the default - the model most people should reach for, most of the time.
Capabilities
From emails and reports to summaries and research notes, Vaani handles the writing work that fills most days. Good structure, clean prose, accurate reasoning - at a speed that keeps up with you.
Vaani writes, reviews, and refactors across any language or framework. Most of what developers ask an AI to do - Vaani does it well. Reach for Kavi only when the problem is genuinely harder than a day's work.
Fast enough to power chat experiences, with 128K of context for conversations that go long. Vaani holds the thread of a discussion without losing nuance or speed.
Pricing
Input
For prompts, context, and documents you send.
Output
For the tokens Vaani generates.
Input costs $0.65 per 1M tokens against Kavi's $0.76, with most of the capability. Built to be the model you leave running.
Choose well
Pick Vaani when the task is normal work - writing, analysis, everyday coding, Q&A. This is most of what you'll do.
Pick Kavi when the problem is unusually hard or the document is unusually long. Vaani will do most jobs; Kavi does the few that need more.
Pick Mira when speed and volume matter more than depth - high-throughput chatbots, classification, real-time autocomplete.