How to Use TokenAir with Kilo Code: Tool Limits
TokenAir can significantly lower Kilo Code model spend when routine coding work uses a lower-cost model without losing tool accuracy. Kilo Code supports custom OpenAI-compatible providers. Agent tasks still depend on verified tool calls and correct model limits.
Compatibility verdict
Limited setup
Kilo Code's custom provider accepts the OpenAI Compatible protocol, Base URL, key, and model list. TokenAir has not yet verified the full file and terminal tool loop, so Agent use remains limited.
A custom endpoint path exists, but important features or request shapes need separate testing.
Before you start
- A current Kilo Code extension or CLI with custom provider support.
- A TokenAir API key and model ID available through /models.
- Verified context and output limits for the selected model before a long Agent task.
Setup and compatibility steps
Step 1
Create a custom provider
Open Kilo Code Settings, go to Providers, and select Custom provider. Give the provider a stable ID such as tokenair.
Step 2
Choose the compatible protocol
Select OpenAI Compatible as the Provider API, then enter the TokenAir Base URL and API key.
Step 3
Select an account-enabled model
Let Kilo fetch models from the endpoint or add the exact ID manually. Do not copy token limits or capability flags from another provider.
Step 4
Test tools before Agent work
Start with a read-only file task. Enable or declare tool_call only after the model completes a real tool call and result round trip.
Kilo Code custom provider fields
Provider ID: tokenair
Provider API: OpenAI Compatible
Base URL: https://api.tokenair.ai/v1
API Key: YOUR_TOKENAIR_API_KEY
Model: YOUR_TOKENAIR_MODEL_IDReplace placeholders with values from your own TokenAir account. Never commit an API key to source control.
How to verify it
- Kilo Code saves the provider and lists the selected model.
- A short chat request completes through TokenAir.
- A file tool call contains valid arguments and returns its result to the model.
- The task reaches a correct final answer without a repeated tool loop.
How to lower cost without hiding quality loss
TokenAir gives the client access to premium and lower-cost model choices through one OpenAI-compatible API. Savings depend on matching each task with a model that still passes review.
- Use one lower-risk coding task to establish tool reliability and total token use.
- Set context and output limits from verified model facts so Kilo can compact at the right time.
- Reserve a stronger model for difficult edits, terminal changes, and final review.
Limits to know before production
- Kilo defaults completely unknown custom models to text input and output only unless capabilities are configured.
- Setting tool_call to true describes an expected capability; it does not make an incompatible endpoint support tools.
- Incorrect context or output limits can cause failed tasks, excessive compaction, or unexpected cost.
FAQ
Can Kilo Code discover TokenAir models automatically?
Kilo Code can fetch available models when the custom endpoint exposes an OpenAI-compatible models route. Use only IDs enabled in your TokenAir account.
Which Provider API should I select?
Select OpenAI Compatible for TokenAir's documented Chat Completions surface. Do not select OpenAI Responses unless TokenAir documents and verifies that path.
Should I set tool_call to true?
Only after a real tool-use smoke test succeeds. The flag tells Kilo what to send, but it cannot supply missing model or endpoint support.
Official sources and related TokenAir docs
Tool settings change between releases. We checked these notes on July 17, 2026. Review the linked tool docs again before a production rollout.
Try one workflow before changing the default.
Use the cost checkup to choose a bounded test, then compare cost per accepted result with your current route.