pub const DEFAULT_MODEL: &str = "@cf/deepseek-ai/deepseek-v4-flash-0731";Expand description
Default inference model - the project’s development default.
DeepSeek V4 Flash is the default because it has demonstrated reliable delivery in the local Hermes tool-use workflow. GLM-5.3 Flash remains available but is experimental pending conformance tests.
Single source of truth: adapters MUST read this (or the policy JSON contract), never duplicate it.