ext/new-api
2
0
Fork 0
mirror of https://github.com/QuantumNous/new-api.git synced 2026-09-12 23:30:35 +00:00
Code Activity

fix(relay): preserve presence/frequency penalty in Responses conversion (#6654)

Browse Source
This commit is contained in:
Uzuki Shion
2026-08-11 13:40:13 +08:00
committed by GitHub
parent 9c97e78ace
commit 253a74dd1b
8 changed files with 172 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
relaykit/dto/openai_request.go
+5
View File
@@ -867,6 +867,11 @@ type OpenAIResponsesRequest struct {
Metadata json.RawMessage `json:"metadata,omitempty"`
Moderation json.RawMessage `json:"moderation,omitempty"`
ParallelToolCalls json.RawMessage `json:"parallel_tool_calls,omitempty"`
// FrequencyPenalty/PresencePenalty are not part of the official OpenAI
// Responses API; they are forwarded verbatim for OpenAI-compatible upstreams
// (e.g. vLLM) that accept them.
FrequencyPenalty json.RawMessage `json:"frequency_penalty,omitempty"`
PresencePenalty json.RawMessage `json:"presence_penalty,omitempty"`
PreviousResponseID string `json:"previous_response_id,omitempty"`
Reasoning *Reasoning `json:"reasoning,omitempty"`
// ServiceTier specifies upstream service level and may affect billing.