Integrations Β· Cline
Cline + every frontier model. One key.
Cline (and its fork Roo Code) supports any OpenAI-compatible provider. Point it at Routify and you get the full DeepSeek / Kimi / GLM / Qwen catalogue plus Claude / GPT / Gemini in the same dropdown β pay one combined balance in USD or CNY.
Setup in 30 seconds
Cline uses an OpenAI-compatible base URL. Override OpenAI Compatible URL and youβre done β no plugin, no extension, no proxy.
π‘ Cline tries diff-and-apply β pair it with deepseek-chat for fast cheap edits, then switch to claude-opus-4-7 only when the diff fails twice. Cline auto-retries with the model in settings β toggle it manually for the second attempt.
# In VS Code β Cline / Roo Code settings:
# API Provider: OpenAI Compatible
# Base URL: https://routify.bytedance.city/v1
# API Key: rtf_xxx_your_routify_key
# Model ID: deepseek-chat
# (or kimi-k2 / glm-4.6 / claude-opus-4-7 / etc.)
# settings.json equivalent:
{
"cline.apiProvider": "openai",
"cline.openAiBaseUrl": "https://routify.bytedance.city/v1",
"cline.openAiModelId": "deepseek-chat",
"cline.openAiApiKey": "rtf_xxx"
}Profiles
Picked for Cline.
Three battle-tested model combos. Pick the profile that matches your priority and copy the model id straight in.
Cheap
DeepSeek V3.2 with Kimi K2.5 fallback for long-context refactors. Cline's diff-and-apply tolerates this combo well β you only feel the difference on agentic tasks > 50 turns.
Cost note
Typical 4-hour Cline session β $0.20.
Primary model
Fallback chain
Smart router auto-fails over to the next id if the primary is over budget or down.
FAQ
Does Cline's diff-and-apply work with DeepSeek?
Yes. Cline uses a search-and-replace patch format that DeepSeek follows correctly ~95% of the time. On the 5% miss, Cline auto-retries β let it.
Will MCP tool servers work?
Yes. MCP servers run on the client side; Cline only needs the model to emit tool_use payloads. Every model on Routify supports tool calls.
What about Roo Code?
Roo Code is a Cline fork β same setup, same model ids. We test both before each release.
How are tokens counted across models?
Each model uses its own tokenizer. Routify shows the actual token count + actual USD cost in dashboard, not an estimate.
Can I cap spend per Cline session?
Set a per-key daily budget in dashboard β Settings β API Keys β Budget. Cline will get an HTTP 429 once the cap is hit; resume tomorrow.