Pricing
Free locally. Pay only when you go to the cloud.
We do not require an API key. The free tier is the full product on local hardware. Plus is for when you want cloud-fallback or Anthropic / OpenAI quality on cloud workloads.
Free
Local-only, no quotas
$0forever
- +All 31 MCP tools
- +All 19 language parsers
- +Ollama models (qwen2.5-coder, nomic-embed-text, qwen2.5-vl)
- +macOS, Windows, Linux, Web
- +Index unlimited local repos
- +MCP server on localhost:19281
- +Telemetry off by default
- +Community support
Most popular
Plus
Bring your own API key
$9/ month
- +Everything in Free
- +Bring-your-own Anthropic / OpenAI / Gemini key
- +Cloud fallback when local model can't fit
- +Larger context windows for huge codebases
- +Provider routing & cost dashboard
- +Priority email support
Team
Shared MCP server for the team
$29/ user / month
- +Everything in Plus
- +Shared MCP server for the team
- +Audit logs
- +SSO (phase 2 promise — see below)
- +Org-wide prompt & tool policies
- +Onboarding session
Self-hosted
For air-gapped orgs
Customtalk to us
- +Run inside your network — no outbound calls
- +Bring your own model server (vLLM, Ollama, custom)
- +Signed builds + checksum manifest
- +Per-deployment licence
- +Direct line to engineering
What you do not pay for
- + Per-tool quotas. There are none.
- + Per-language unlock fees. All 19 included on free.
- + Index size limits. Your disk is the limit.
- + Telemetry surcharges. We do not sell your usage.
- + "AI credits". You bring (or run) the model.
- + Forced cloud upload of your source code. Ever.
FAQ
Do I need an API key to use Tanvrit AI?+
No. The free tier is the full product running on your local hardware via Ollama. Plus is for when you want cloud-fallback (Anthropic, OpenAI, Gemini) for bigger codebases or higher-quality generation.
Which models work?+
Locally: any Ollama-compatible model. We recommend qwen2.5-coder for planning, nomic-embed-text for embeddings, and qwen2.5-vl for vision (with Tesseract OCR fallback). Remotely (Plus/Team): Claude, GPT-4 family, Gemini, DeepSeek, Groq, Mistral.
What hardware do I need?+
8 GB RAM is enough for small Ollama models. 16 GB is comfortable for qwen2.5-coder 7B. 32 GB+ unlocks the larger models. Apple Silicon and modern AMD/Intel CPUs both work; a discrete GPU helps but isn't required.
Is telemetry on?+
Off by default. If you opt in we collect anonymised counters (active platform, tool-use counts, error class) — never paths, URLs, emails, or code. PII guard rejects anything that looks like one. Endpoint is documented and can be blocked at the network layer.
Where does my data live?+
On the free tier, everything stays on your machine: the SQLDelight index, embeddings, MCP bearer token. Plus and Team only send the chunks your agent explicitly requests, and only to the AI provider you configured.
Is there a free trial for Plus?+
The free tier is the trial. Start there, upgrade when you actually need cloud-fallback. We don't believe in fake limits to push upgrades.
Cancellation and refunds?+
Plus is month-to-month, cancel any time from the billing page. Team is annual with a 30-day refund window. Self-hosted licences are negotiated per deployment.