Affected by 
Documentation
¶
Overview ¶
Package reasoning re-exports the pure model-name effort-suffix helpers, which moved to the conversion kit (relaykit/relayconvert/reasoning) as part of the relaykit extraction. Host code keeps importing this path unchanged.
Index ¶
- Variables
- func BaseModelName(modelName string) string
- func CanonicalBillingModelNames(modelName string) []string
- func ParseLegacyModelSuffix(modelName string, allowClaudeThinkingAlias bool, allowGeminiThinkingAlias bool) (string, kitreasoning.Intent, bool, error)
- func ParseOpenAIReasoningEffortFromModelSuffix(modelName string) (string, string)
Constants ¶
This section is empty.
Variables ¶
var ( EffortSuffixes = kitreasoning.EffortSuffixes OpenAIEffortSuffixes = kitreasoning.OpenAIEffortSuffixes DeepSeekV4EffortSuffixes = kitreasoning.DeepSeekV4EffortSuffixes )
var ( TrimEffortSuffixWithSuffixes = kitreasoning.TrimEffortSuffixWithSuffixes ParseDeepSeekV4ThinkingSuffix = kitreasoning.ParseDeepSeekV4ThinkingSuffix TrimGeminiThinkingSuffix = kitreasoning.TrimGeminiThinkingSuffix )
Functions ¶
func BaseModelName ¶
BaseModelName strips explicit model modifiers and any enabled legacy alias. Names on the thinking-suffix blacklist stay verbatim, including @ tails. Malformed legacy aliases stay intact so request conversion can report the precise validation error later.
func CanonicalBillingModelNames ¶
CanonicalBillingModelNames returns specificity-descending canonical billing name candidates (without the raw request name or the bare base). Explicit @ modifiers and legacy aliases normalize through the same Intent, so order, duplicates, and case do not matter. Temperature and topp never appear.
func ParseLegacyModelSuffix ¶
func ParseLegacyModelSuffix(modelName string, allowClaudeThinkingAlias bool, allowGeminiThinkingAlias bool) (string, kitreasoning.Intent, bool, error)
ParseLegacyModelSuffix parses the old naked aliases only for positively matched GPT/o-series, Claude, and Gemini model families. The provider prefix before the final path segment is kept opaque.
func ParseOpenAIReasoningEffortFromModelSuffix ¶ added in v0.13.1
ParseOpenAIReasoningEffortFromModelSuffix applies RelayKit's positive family whitelist and the host EffortTailModelIDs escape hatch so real model IDs such as gpt-5.1-codex-max remain opaque.
Types ¶
This section is empty.
Source Files