Cruxy
Talk to sales
Join waitlist
Join waitlist
ProductsCruxyCruxy CodeCruxy CoworkCruxy BenchCruxy Guard

Cruxy — India's AI, one API call away.

© 2026 Cruxy. Built by the team behind mycrux..

cruxy@mycrux.in
PrivacyTermsUsage
Model › Kavi

Kavi. Our most capable model.

Built for the problems that need depth. Kavi thinks longer, reasons further, and handles the work where being right matters more than being fast.

Try KaviAPI reference →
250K context·16K max output·Text + vision·26+ languages·₹850/1M input · ₹4,250/1M output

Capabilities

Kavi is for deep work.

Deep reasoning

Kavi handles multi-step problems where getting each step right affects the final answer. Complex analysis, planning, and tradeoff evaluation are where Kavi stands apart from faster models.

Long-document understanding

With 250K tokens of context - roughly 500 pages - Kavi reads entire research papers, codebases, or legal documents in one pass. Ask questions, summarize, cross-reference, without chunking.

Nuanced writing and code

For work that needs voice, judgment, or architectural thinking, Kavi writes with awareness of context and constraint. Long reports, refactors, complex prompts - Kavi holds the thread.

Pricing

Priced for depth.

Input

₹850per 1M tokens

For the tokens you send to Kavi - prompts, documents, context. $10 per 1M in USD.

Output

₹4,250per 1M tokens

For the tokens Kavi generates in response. $50 per 1M in USD.

30% lower than comparable models from US labs. No hidden fees. Volume discounts available.

See full pricing →

Choose well

When to pick Kavi over Vaani or Mira.

Pick Kavi when the answer must be right, the document is long, or the reasoning is layered.

Pick Vaani when the task is everyday - writing, summarizing, answering questions, coding most features. Vaani is the default.

Pick Mira when speed matters more than depth - chatbots, classification, real-time responses.

Compare all three →

Ready for depth?

Try KaviAPI reference →