Knowledge Base Sections ▾
Navigation
▸ Start here By rolesCategories
- OpenCode: free models, providers and API key
- Cursor + Gonka AI - cheap LLM for coding
- Claude Code + Gonka AI - LLM for the terminal
- DeepSeek Harness: Your Own Provider via JoinGonka Gateway
- Codex CLI: your own key instead of a subscription
- Pi + JoinGonka Gateway — terminal agent on cheap inference
- omp (Oh My Pi) + JoinGonka Gateway: an agent with model roles
- Zoo Code + JoinGonka Gateway: Migrating from Roo Code to Gonka models
- Kilo Code + Gonka AI — AI-Agent in VS Code
- ZCode: GLM Coding Plan tiers, prices and limits
- Hermes Agent + DeepSeek on the Gonka network — an autonomous agent for pennies
- OpenClaw + Gonka AI - affordable AI agents
- Continue.dev + Gonka AI - AI for VS Code/JetBrains
- Cline + Gonka AI - AI agent in VS Code
- Aider + Gonka AI - pair programming with AI
- LangChain + Gonka AI - AI applications for pennies
- n8n + Gonka AI - automation with cheap AI
- Open WebUI + Gonka AI - your own ChatGPT
- LibreChat + Gonka AI — open-source ChatGPT
- Roo Code + Gonka AI — Autonomous AI Agent in VS Code
- LlamaIndex + Gonka AI — RAG applications for pennies
- PydanticAI + Gonka — typed AI agents for pennies
- Vercel AI SDK + Gonka AI — AI applications in TypeScript for pennies
- TanStack AI + Gonka — AI applications in TypeScript for pennies
- API quick start — curl, Python, TypeScript
- JoinGonka Gateway — a full overview
- Management Keys — SaaS on Gonka
- Cheapest AI API: Provider Comparison 2026
- How to buy AI tokens and an API key: 3 methods in 2026
- Free AI API: tokens on sign-up and for referrals
- Cursor Pro request limit reached — breakdown and cheaper alternative
- Cheaper Claude Code: GLM and DeepSeek instead of Claude
- Claude Code + DeepSeek V4 Flash: How to Connect It and What a Task Costs
- Cline is burning money — why the agent spends so much
- OpenClaw is expensive — why the agent burns through tokens and how to save
- OpenRouter alternative: the same models, many times cheaper
- Best AI model for coding in 2026: comparison and prices
- Cheap alternative to GitHub Copilot without limits
- A cheap Windsurf alternative without credits or limits
- The cheapest API for AI agents in 2026
- JetBrains IDE + JoinGonka Gateway — your own endpoint instead of credits
- GitHub Copilot BYOK — own models instead of quotas
- Zed + JoinGonka Gateway — cheap inference in your editor
- MiniMax Code: MiniMax agent with your own key via Gonka
- Warp + JoinGonka Gateway — terminal agent on your own endpoint
- Trae + JoinGonka Gateway — Gonka network models in AI-IDE
- Cherry Studio + JoinGonka Gateway — desktop AI client
- OpenHands + JoinGonka Gateway: agent on your own endpoint
- Qwen Code after the closure of qwen-oauth: working via JoinGonka Gateway
- Goose + JoinGonka Gateway: your own provider and key in the keyring
- Crush + JoinGonka Gateway: Charm agent on Gonka network models
- Kimi Code CLI: Moonshot AI agent on your key via Gonka
- Factory Droid + JoinGonka Gateway: BYOK on Gonka network models
- MiMo Code + JoinGonka Gateway: Xiaomi agent on Gonka network models
Tools
Roo Code + Gonka AI — Autonomous AI Agent in VS Code
Important: The Roo Code extension was discontinued on May 15, 2026. The repository has been archived, and there will be no further updates. Development is being continued by Zoo Code — a fork started by the Roo Code community: the provider configuration is identical, and for moving, there is a migration guide and our Zoo Code guide. Other open agents for VS Code include Kilo Code and Cline, from which Roo Code was originally forked. An already installed extension continues to work, and the instructions below are still relevant for it.
Roo Code is an autonomous AI agent for VS Code: it reads and edits files, executes terminal commands, interacts with the browser, and completes multi-step development tasks. It is a fork of Cline that went further: customizable modes (Architect, Code, Ask, Debug, and custom ones), project-specific instructions, and a flexible model configuration panel. In essence, it is an entire team of AI specialists inside your editor.
The main problem with such agents is token consumption. For a single task, Roo Code processes file contexts, command outputs, and tool results — amounting to tens of millions of tokens. At Anthropic's prices ($3-15 per 1M), a single full session turns into $30-1500. For daily work, this is unrealistic.
JoinGonka Gateway reduces the cost by thousands of times: the same session will cost $0.01-1.00. The Gateway resells inference from the decentralized Gonka network and supports both the OpenAI format ("OpenAI Compatible" provider) and the Anthropic format (endpoint /v1/messages) — Roo Code connects via either. This transforms Roo Code from an expensive demo into an everyday working tool.
Step 1: Install Roo Code and get a key
Installing Roo Code: in VS Code, open Extensions (Ctrl/Cmd+Shift+X), search for "Roo Code" and click Install. After installation, a kangaroo icon for Roo Code will appear in the sidebar. The extension is still in the Marketplace, but its last version — 3.54.0 from May 15, 2026 — will remain the final one.
JoinGonka API key: if you don't have a key yet — sign up at gate.joingonka.ai/register, get 3M free tokens, and create a key with the jg- prefix in the Dashboard.
Step 2: Configure Roo Code (OpenAI Compatible)
The easiest way is to set it up with the one-command installer. For CLI/file tools it will configure the JoinGonka provider itself (baseUrl, models), and for UI tools it will show ready-to-paste values:
npx @joingonka/setup --tool rooThe command first reminds you to close Roo Code and tells you where to go (--tool zoo, --tool cline, --tool kilo), then prints the values for those who still have the extension installed — along with the real context window and the model's response ceiling. The installer will ask for your jg-… key and finish with a live check via a request to the gateway. Below is the manual method if you prefer to configure it yourself.
Set up manually (Plan B)
Open the Roo Code panel and go to the settings (gear icon). In the provider section, set:
- API Provider — select
OpenAI Compatible. - Base URL —
https://gate.joingonka.ai/v1 - API Key —
jg-your-key - Model (Model ID) —
MiniMaxAI/MiniMax-M2.7
Below, in the Model Configuration block, you can fine-tune the model parameters manually:
- Context Window —
180000(180K tokens; for DeepSeek V4 Flash — 380000). - Max Output Tokens —
8192for MiniMax M2.7 and GLM-5.3 Flash,32768for DeepSeek V4 Flash. - Input Price / Output Price — you can enter the real rates (≈$0.0062 per 1M input, output ×3) so that Roo Code calculates task costs correctly right in the interface.
Important about tool calling: Roo Code uses only native tool calling — there is no XML fallback here, unlike in older agents. So the model must support function calling. The network's models via our Gateway support native tool calling — Roo Code works with them out of the box.
Check: in the Roo Code chat, type “Create a hello.py file with a function that prints Hello World”. The agent will offer to create the file and show a diff for approval.
Comparing Agent Session Costs
Roo Code is an agentic tool: it does not just reply with one message, but executes a task — it reads files, writes code, runs tests, and fixes bugs. Each action is a model call. Let's compare the costs of typical sessions:
| Task | Tokens | Anthropic Claude | OpenAI GPT | JoinGonka Gonka |
|---|---|---|---|---|
| Simple bug fix | ~5M | $15 — $75 | $12 — $50 | $0.048 |
| New feature (2-3 files) | ~20M | $60 — $300 | $50 — $200 | $0.19 |
| Module refactoring | ~50M | $150 — $750 | $125 — $500 | $0.48 |
| Full development session (4h) | ~100M | $300 — $1,500 | $250 — $1,000 | $0.96 |
With JoinGonka Gateway, Roo Code becomes an everyday tool — you can run it for every ticket, every bug, every feature, without worrying about the bill. 3M free tokens at the start allow you to try the setup without a card.
Model parameters: MiniMax M2.7 — 180K window and up to 8192 response tokens, DeepSeek V4 Flash — 380K and up to 32768, GLM-5.3 Flash — 390K and up to 8192. If max_tokens for a non-stream request is not specified, the Gateway defaults to returning up to 1500 tokens (up to 3000 for GLM-5.3 Flash). Roo Code automatically breaks long generations into steps.
How Roo Code differs from Cline: modes and Anthropic format
Roo Code is a fork of Cline, but with notable differences that change the workflow:
- Modes: switch the agent's role to match the task — Architect for planning, Code for writing code, Ask for project questions, Debug for finding bugs. You can create your own modes with separate instructions and a set of allowed tools.
- Native tool calling only: unlike agents with text/XML parsing, Roo Code sends tools via the native OpenAI
toolsschema and receives calls as separate events — lower latency and more reliable results. The flip side: the model must support function calling (the network's models via our Gateway do). - Custom instructions and profiles: per-project rules (for example, via a rules file in the repository) and multiple saved provider configurations you can switch between easily.
Connecting via the Anthropic format. If you want to use our /v1/messages endpoint instead of the OpenAI format: in settings, select API Provider → Anthropic, check "Use custom base URL" and specify https://gate.joingonka.ai, paste the same jg-your-key key into the Anthropic API Key field, then set the model name (for example MiniMaxAI/MiniMax-M2.7). The key and balance are shared across both formats, and native tool_use in Anthropic mode also goes through the Gateway.