Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 14 additions & 3 deletions src/api/providers/openai-native.ts
Original file line number Diff line number Diff line change
Expand Up @@ -41,7 +41,17 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
}

if (model.id.startsWith("o3-mini")) {
yield* this.handleO3FamilyMessage(model, systemPrompt, messages)
yield* this.handleReasonerMessage(model, "o3-mini", systemPrompt, messages)
return
}

if (model.id.startsWith("o3")) {
yield* this.handleReasonerMessage(model, "o3", systemPrompt, messages)
return
}

if (model.id.startsWith("o4-mini")) {
yield* this.handleReasonerMessage(model, "o4-mini", systemPrompt, messages)
return
}

Expand Down Expand Up @@ -72,13 +82,14 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
yield* this.handleStreamResponse(response, model)
}

private async *handleO3FamilyMessage(
private async *handleReasonerMessage(
model: OpenAiNativeModel,
family: "o3-mini" | "o3" | "o4-mini",
systemPrompt: string,
messages: Anthropic.Messages.MessageParam[],
): ApiStream {
const stream = await this.client.chat.completions.create({
model: "o3-mini",
model: family,
messages: [
{
role: "developer",
Expand Down
21 changes: 21 additions & 0 deletions src/shared/api.ts
Original file line number Diff line number Diff line change
Expand Up @@ -786,6 +786,27 @@ export const openAiNativeModels = {
inputPrice: 10.0,
outputPrice: 40.0,
cacheReadsPrice: 2.5,
reasoningEffort: "medium",
},
"o3-high": {
maxTokens: 100_000,
contextWindow: 200_000,
supportsImages: true,
supportsPromptCache: true,
inputPrice: 10.0,
outputPrice: 40.0,
cacheReadsPrice: 2.5,
reasoningEffort: "high",
},
"o3-low": {
maxTokens: 100_000,
contextWindow: 200_000,
supportsImages: true,
supportsPromptCache: true,
inputPrice: 10.0,
outputPrice: 40.0,
cacheReadsPrice: 2.5,
reasoningEffort: "low",
},
"o4-mini": {
maxTokens: 100_000,
Expand Down