Pricing

Simple, sovereign pricing

Invite-only preview today. Team and Enterprise Dedicated as we scale. Usage is metered in credits per developer — and every plan includes the independent referee.

Most popular

Pilot

Invite
Free during the invite-only preview. Real scarcity — limited seats per machine.
  • Full agentic workflow in VS Code
  • Independent verification (referee) built in
  • Data plane in India
  • One transferable invite for a friend

Team

Waitlist
Pooled seats for startups and teams. Per-developer metering, org console.
  • Per-dev keys & metering
  • Pooled capacity, fair-use
  • Org console + limits
  • SSO

Enterprise Dedicated

₹ on request
Your own dedicated serve. BFSI/Defence-ready. Annual commit.
  • Dedicated GPU capacity — no sharing
  • In your VPC or air-gapped
  • Independent verification + audit trail
  • Uptime monitoring + SLA
  • Up to 50 developers

Max add-on

Metered
Frontier-class node with 160K context, provisioned on demand.
  • 160K context window
  • Dedicated 4-GPU node
  • Billed per GPU-hour
  • For the largest codebases

Pricing is finalised with each customer in ₹ (INR-first, GST-compliant). Usage metered in credits (tokens × model weight). Dedicated capacity — your own reserved GPU serve, in-VPC or air-gapped — is the flagship enterprise offering: talk to us.

Taksha vs. foreign coding clouds

Same IDE workflow. One difference that matters to a regulated buyer: where the code goes, and who checks the claims.

Foreign coding cloudsTaksha
Where your source code goesUS / foreign cloudIndia — or never leaves your VPC
Model ownershipRented frontier APIOwned & operated model family
Independent verification of resultsModel grades its own workReferee stamp — a check the model never sees
Publishes failure benchmarksNoYes — including cells we lose
Fine-tuned on your estateNoYes (enterprise)
Air-gapped deploymentYes

The Taksha family — pick the right tool

Every model is served on our own sovereign infrastructure, in your region. Heavier models cost more credits (credits = tokens × the model’s weight), so reach for the lightest one that does the job.

Pro flagship

taksha-pro · the everyday default
5× credits
The general model most teams live in
  • Full agentic loop: edits, runs, verifies, finishes
  • Feature work, refactors, debugging, review
  • Any OpenAI-compatible editor or SDK
  • 45/45 on our graded agentic battery — zero phantom completions
The middle of the capability ladder (Fast → Pro → Max) and the right default for almost everything. Measured at parity with a leading frontier model on the same graded tasks, from a sovereign box.
Unlocked from Team · API model id taksha-pro

Max Enterprise · on-demand

taksha-max · frontier-class
20× credits
Frontier-class + 160K context
  • 45/45 on our graded agentic battery — zero phantoms
  • The largest codebases and long multi-file context
  • Hardest architectural problems
  • Provisioned on request in about 15 minutes
A dedicated multi-GPU node provisioned on demand. Same sovereign perimeter, same independent verification.
Unlocked from Enterprise · API model id taksha-max

Pariksha

taksha-pariksha · testing & QA
5× credits
परीक्षा — tests that actually catch bugs
  • Unit · integration · functional (pytest, JUnit, Jest, NUnit, Go)
  • E2E scripting: Playwright, Selenium, Appium
  • Performance scripts: k6, JMeter, Gatling with stated thresholds
  • Mutation-checked: it breaks your code to prove the suite goes red
The testing specialist. Its doctrine forbids the industry default — suites that stay green on broken code. Gate battery (31 Jul): measured at parity with a leading frontier model on mutation-kill rate — 89% of seeded defects caught, zero invalid suites — from a sovereign box.
Unlocked from Pilot · API model id taksha-pariksha

Kriti

taksha-kriti · UI craft
5× credits
कृति — interfaces that actually work, available now
  • Components that are wired, not just written
  • React · Angular · Vue · vanilla
  • Graded against a real DOM, not eyeballed
  • Exact ids, real event wiring, honest starting state
The UI specialist. Gate battery (3 Aug): +15 points over the base engine across 8 interactive UI task classes, n=128 per arm, zero classes regressed — measured by driving the rendered DOM, not by reading the code.
Unlocked from Pilot · API model id taksha-kriti

Yantra coming soon

taksha-yantra · for teams building AI
5× credits
यन्त्र — for teams building agentic AI, in development
  • Agent frameworks: LangChain/LangGraph, OpenAI + Google agent SDKs
  • Tool protocols: MCP servers and clients
  • Retrieval pipelines, eval harnesses, guardrails
  • For teams whose product IS the agent
The agentic-AI specialist. In development — it ships only when it passes its own release gate, same as every tier here. Today taksha-yantra routes to the Pro arm, so existing installs keep working unchanged.
In development · join the waitlist

Adhunik

taksha-adhunik · modernization
5× credits
आधुनिक — legacy to modern, available now
  • COBOL · VB.NET · mainframe estates
  • Proven: COBOL→Spring with exact-arithmetic verification
  • VB.NET→C# triple-verified builds
  • Referee-stamped at every step
The modernization identity of our fine-tuned engine — trained on real modernization estates. For India's legacy backlog, with receipts.
Unlocked from Pilot · API model id taksha-adhunik
The Measured Router

taksha-auto — routed to the model that measures best on your task class

receipt in every response

Can't decide? Don't. Each request is classified and routed by a public per-class table — every cell is pass/n from execution-graded battery ledgers, never hand-edited — and every response carries x-taksha-router-receipt saying what routed where, and why. taksha-auto-quality and taksha-auto-cost tune the trade-off.

Which should I use?
Multi-step agentic task
Yantra. The default — it finishes and verifies.
Sharpest single-shot generation
Pro. Fine-tuned for output quality.
Huge codebase / 160K context
Max (Enterprise · on-demand).
Not sure?
Start with Yantra and set taksha.verifyCommand — let the referee keep score.
Route it per request
taksha-auto. The Measured Router picks by the graded table, receipt in every response.
Using it from your editor

Any OpenAI-compatible client (Continue, Cline, Zed, Cursor custom base URL). Set the base URL + your key, then pick a model by its id — the proxy meters credits and enforces your plan automatically.

OPENAI_BASE_URL=https://app.agentanywhere.ai/agents/coder/v1 OPENAI_API_KEY=sk-sov-live-… # from your dashboard model: sovereign-coder-fast | sovereign-coder-open | sovereign-coder | sovereign-coder-pro | sovereign-coder-ultra

Response headers X-Credits-Used / X-Credits-Remaining show your balance on every call. Locked models return a clear upgrade hint.

Sovereignty, not just privacy

Region-pinned. In-VPC. Owned model layer.

Every completion is served from a model we own, on hardware in your jurisdiction. For regulated buyers, that's the difference between a policy promise and a network fact.

Book an enterprise pilot