One account, one session history, every major IDE. Move between editors during the week and the agent picks up where you left it instead of starting over.
VS Code
Cursor
JetBrains
Neovim
Tok Net Code
Like Claude Code and Codex. Half the price.
The coding agent you already know: it reads the code, plans the change, edits across files and runs the tests, right inside your IDE. What is different is the bill. Every model from every lab instead of one, and 50% cheaper than either for the same amount of coding.
Claude · GPT · Gemini · 200+ models
€13 or €120 per month
Claude
GPT
Gemini
Llama
+200
The checkout test is flaky. Find out why and fix it without touching the retry logic.
The poller shares a clock with the fixture, so a slow CI box lands one tick early. Patched the fixture, left the retry path alone, suite is green.
every IDE
every model
half the bill
What it is
A coding agent, made for coding.
It plans, edits across files, runs your tests and tells you what it changed. The part you cannot get anywhere else is what sits underneath it.
Claude, GPT, Gemini, Llama and 200 more behind one key. Swap the model in the middle of a task and the agent keeps the context, so you can reach for the strong one only where it earns its price.
Claude
GPT
Gemini
Llama
+200
Every request goes through the Tok Net AI gateway first, which compresses the prompt, answers from cache where it can and routes the rest to the cheapest model that clears your quality floor. Same agent, same amount of coding, half the bill.
compress
cache
route
Plans
Two subscriptions. Both cheaper than the alternative.
Flat monthly price, every model included. No seat minimums and no per-token surprises.
€13 / month
For one developer who codes with an agent most days. Roughly what a €25 plan elsewhere gives you, for half.
- Every model, in every major IDE
- Switch model mid-task, context carries over
- Compression and caching on every request
- A quality floor routing never trades below
€120 / month
For all-day agent use, where the agent is running as often as you are typing. The ceiling of a €200-plus plan elsewhere.
- Everything in Pro
- The highest usage ceiling we offer
- Shared team memory and prompt library
- Priority routing when providers are busy
Comparison
Tok Net Code, Claude Code and Codex.
The three agents people put side by side. Here is where they actually differ.
Coding agents compared scroll sideways
| Tok Net Code | Claude Code | Codex | |
|---|---|---|---|
| Models | All of them: Claude, GPT, Gemini, Llama and 200+ more | All Claude models | All OpenAI models |
| Usage limits | Higher than both, on the same money | 5-hour and weekly caps | 5-hour and weekly caps |
| Editors | Every major IDE | VS Code, JetBrains | VS Code |
| Switch model mid-task | Yes, context carries over | Within Claude only | Within OpenAI only |
| Lock-in | None; one key, swap providers whenever | Anthropic account | OpenAI account |
| Price | €13 or €120 per month | $20–$200 per user / month | $20–$200 per user / month |
Competitor plans and limits as published by Anthropic and OpenAI, July 2026.
FAQ
Before you ask.
How can it be half the price for the same work?
Because the agent does not talk to the provider directly. Every request goes through the Tok Net AI gateway, which compresses the prompt down to what the model actually needs and answers repeat work from cache. See exactly how that works in token and cost optimization.
Which editors does it run in?
VS Code, Cursor, JetBrains and Neovim, all on the same account and the same session history. Move between editors during the week and the agent picks up the thread instead of starting over.
Can I keep using Claude or GPT specifically?
Yes. Pin the model per project or per task and the agent will use only that one. The difference is that you can also change your mind halfway through a task, and the context comes with you instead of starting over.
Do you train on our code?
No. Your prompts and your code stay yours, we do not train on them and we do not sell them. The security page has the detail.
Tok Net Code
Same coding. Half the bill.
The beta opens soon. Join the waitlist and we will hand you a key before it goes public.