kunthawat/moreminimore-vibe
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Lower o3/o4 mini max output tokens to maximize input tokens (#424)

Browse Source
This commit is contained in:
Will Chen
2025-06-17 10:31:24 -07:00
committed by GitHub
parent d69284e3b0
commit 5694934183
1 changed files with 6 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
src/ipc/shared/language_model_helpers.ts
View File

@@ -44,7 +44,8 @@ export const MODEL_OPTIONS: Record<string, ModelOption[]> = {
name: "o3-mini", name: "o3-mini",
displayName: "o3 mini", displayName: "o3 mini",
description: "Reasoning model", description: "Reasoning model",
maxOutputTokens: 100_000, // See o4-mini comment below for why we set this to 32k
maxOutputTokens: 32_000,
contextWindow: 200_000, contextWindow: 200_000,
}, },
// https://platform.openai.com/docs/models/o4-mini // https://platform.openai.com/docs/models/o4-mini
@@ -52,7 +53,10 @@ export const MODEL_OPTIONS: Record<string, ModelOption[]> = {
name: "o4-mini", name: "o4-mini",
displayName: "o4 mini", displayName: "o4 mini",
description: "Reasoning model", description: "Reasoning model",
maxOutputTokens: 100_000, // Technically the max output tokens is 100k, *however* if the user has a lot of input tokens,
// then setting a high max output token will cause the request to fail because
// the max output tokens is *included* in the context window limit.
maxOutputTokens: 32_000,
contextWindow: 200_000, contextWindow: 200_000,
}, },
], ],