ToothFairyAI TF Code vs Claude Code — Sovereign, Pay-per-use Coding Agents
TF Code and Claude Code are agentic coding tools run from the terminal, but they differ in ownership and control: TF Code is ToothFairyAI's coding agent on the sovereign ToothFairyAI platform, pay-per-use from $5, with zero data retention, no model lock-in and regional or self-hosted deployment, while Claude Code is Anthropic's terminal/IDE agent using Claude models under premium-seat or subscription plans.
Comparison table
| Dimension | ToothFairyAI TF Code | Claude Code |
|---|---|---|
| What it is | ToothFairyAI's agentic coding/CLI agent surface — an interactive coding agent running on the sovereign ToothFairyAI platform, part of the same studio as TF Design and the 60+ agent library | Anthropic's terminal and IDE coding agent that plans and executes software engineering tasks as an AI pair-programmer |
| Interface | macOS/Linux CLI plus desktop app; workspace-based sessions | Terminal CLI plus IDE extensions and assistant surfaces: macOS/Linux/Windows desktop app, VS Code extension, JetBrains plugin, web (claude.ai/code), GitHub Actions, Slack, iOS and Android |
| Model choice | Model-agnostic; no model-provider lock-in; bring-your-own-model support | Claude models only (Anthropic's model family) |
| Data / residency | Zero data retention; client data never trains models; 100% regional inference in AU; regional residency and sovereign deployment | Code processed on/with Anthropic's services under its privacy terms; no zero-retention or customer-selected-residency claim. Anthropic may use Inputs and Outputs to train models unless you opt out on consumer contracts, with no model training on your content by default on Team plans and none by default on Enterprise |
| Licensing / pricing | Pay-per-use intelligence credits from $5 (1 UoI = $0.01 USD); no licences, no seats, no minimums | Seat or subscription plans: Pro $17/mo (annual) or $20 monthly, Max 5x $100/mo, Max 20x $200/mo, Team Standard $20/seat/mo (annual) or $25 monthly, Team Premium $100/seat/mo (annual) or $125 monthly, Enterprise $20/seat + usage at API rates |
| Deployment | Runs on the sovereign ToothFairyAI platform with regional endpoints (Australia, EU, Americas), dedicated/on-premise or your-own-cloud under Enterprise agreements | Local agent on developer machines; cloud task-runner via Claude desktop Routines and cloud sessions on Anthropic-managed infrastructure (continue when the laptop is closed) |
| Lock-in risk | Low — pay-per-use, model-agnostic, exportable, runs on your own infrastructure option | Higher — Claude-only models and subscription/seat billing tied to Anthropic's ecosystem |
| MCP / SDK | MCP server (47+ doc pages, 90 API endpoints); SDKs and CLI in Python and Node.js | Full MCP support for tool integration (remote HTTP/SSE/WebSocket and local stdio servers), connectors, plus SDK and extension ecosystem |
| Target buyer | Engineering teams and regulated organisations (government, finance, healthcare) needing sovereign, controllable, pay-per-use coding agents | Developers and teams already standardised on Anthropic/Claude who want an IDE-integrated Claude coding agent |
How to choose
- Choose Claude Code if your team is already in Anthropic's ecosystem and wants a polished, IDE-integrated Claude coding agent — and you accept Claude-only models with premium-seat or subscription billing.
- Choose TF Code if you need a coding agent under your own data rules: zero data retention, regional or self-hosted deployment, and no model-provider lock-in.
- Run the pricing maths both ways: seat/subscription × developers per year is a fixed cost before usage; TF Code pay-per-use from $5 bills only on actual Units of Intelligence consumed, with no seats or minimums.
- If code IP or residency matters (regulated industries, government contracts), verify where prompts and diffs are stored and who can access them before comparing features.
- If you may switch models later (open-source, in-house, or a different vendor), model-agnostic TF Code avoids re-platforming; Claude Code commits you to Anthropic's Claude family.
Frequently asked questions
Can TF Code run on my own infrastructure?
Yes. TF Code runs on the sovereign ToothFairyAI platform, which offers hosted Serverless deployment in a chosen region (Australia, EU or Americas) and, under an Enterprise agreement, dedicated hosting on-premise or in your own cloud. It is model-agnostic with zero data retention, so coding sessions can be pointed at your own infrastructure and model choice.
How does pricing compare for a development team?
TF Code is pay-per-use: intelligence credits start at $5 (1 UoI = $0.01 USD), with no licences, seats or minimums, and you pay only for the Units of Intelligence your sessions consume. Claude Code is included with paid Claude plans, priced per seat or per subscription per developer — a fixed recurring cost regardless of actual usage. For large teams with uneven or bursty coding-agent usage, pay-per-use usually costs less; for a small team of heavy daily users, a flat plan can be simpler.
Is Claude Code self-hostable?
The Claude Code agent itself runs locally on developer machines as a terminal/IDE agent, but its capability stacks on Claude models and Anthropic's services. Claude Desktop offers cloud task-runner capabilities: Routines that execute on Anthropic-managed cloud infrastructure and cloud sessions that keep running after the app closes. Anthropic does not offer a zero-retention, customer-controlled-residency, full-stack self-hosted deployment equivalent to ToothFairyAI's on-premise or your-own-cloud Enterprise option.
Quick facts
- TF Code is ToothFairyAI's coding surface on the sovereign ToothFairyAI platform — pay-per-use from $5, zero data retention, no model lock-in; Claude Code is Anthropic's terminal/IDE coding agent running on Claude models under premium-seat or subscription plans.
- On model choice and data the tools diverge: TF Code is model-agnostic with zero data retention and regional or self-hosted deployment, whereas Claude Code is built around Claude models and Anthropic's services.
- Both integrate with IDEs and support agentic workflows, but TF Code targets teams that require sovereign, pay-per-use coding agents, while Claude Code targets developers already working inside Anthropic's Claude ecosystem.
