Included models
Paid plans include Skyflo-managed model access, subject to their allowance and your managed-content choice. Free uses your own model keys or coding-agent access instead.
| Model | Core | Pro | Max |
|---|---|---|---|
| Luna | Available | Available | Available |
| Gemini 3.8 Flash | Available | Available | Available |
| Terra | Available | Available | Available |
| Sol | Not included | Available | Available |
| Astra | Not included | Choose explicitly | Available |
Z.ai models are no longer included in the managed catalogue. Supported Z.ai access through your own key is a separate path.
Auto and Efficient
Auto routes among eligible models for the work. Core can route among Luna, Gemini 3.8 Flash and Terra. Pro can also route to Sol; choose Astra explicitly on Pro. Max can reach Astra automatically when justified. Efficient chooses the efficient route. Choosing a named model pins that choice rather than requesting Auto routing.
Model availability can change with your plan, remaining allowance, provider readiness and account settings. The live picker is the authority for a new turn.
Effort and context
Effort controls how much reasoning a compatible model may use. Core offers up to Balanced, Pro up to Deep and Max up to Max. A model can support fewer effort levels than the plan permits.
The managed input limits shown by the product are 64,000 tokens on Core, 128,000 on Pro and 272,000 on Max per call. These are plan limits, not a claim about the provider model's full context capacity. Conversation history, instructions and tool results share the available context.
External agents
A coding agent's own model list comes from its configuration and account. A model available in Codex or Pi does not thereby become an included Skyflo model.