fix:fix gemini Maximum output tokens value #661
Conversation
WalkthroughThe update modifies the Changes
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~2 minutes Poem
Note ⚡️ Unit Test Generation is now available in beta!Learn more here, or try it out under "Finishing Touches" below. ✨ Finishing Touches
🧪 Generate unit tests
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments. CodeRabbit Commands (Invoked using PR comments)
Other keywords and placeholders
CodeRabbit Configuration File (
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
src/main/presenter/configPresenter/modelDefaultSettings.ts (1)
367-369: Flash-Lite values now diverge from Pro/Flash – intentional?
maxTokensis raised to 65 536 while the two siblings were lowered to 65 535. If this is intentional (different model family) please add a short inline comment citing the source, otherwise align the three for clarity.
Also notecontextLengthis bumped to 1 048 576; if that figure comes from new docs, consider extracting it into a typed constant so future API updates don’t require hunting through this massive literal array.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
src/main/presenter/configPresenter/modelDefaultSettings.ts(3 hunks)
🧰 Additional context used
📓 Path-based instructions (8)
**/*.{ts,tsx,js,jsx,vue}
📄 CodeRabbit Inference Engine (CLAUDE.md)
Use English for logs and comments
Files:
src/main/presenter/configPresenter/modelDefaultSettings.ts
**/*.{ts,tsx}
📄 CodeRabbit Inference Engine (CLAUDE.md)
Strict type checking enabled for TypeScript
**/*.{ts,tsx}: 始终使用 try-catch 处理可能的错误
提供有意义的错误信息
记录详细的错误日志
优雅降级处理
日志应包含时间戳、日志级别、错误代码、错误描述、堆栈跟踪（如适用）、相关上下文信息
日志级别应包括 ERROR、WARN、INFO、DEBUG
不要吞掉错误
提供用户友好的错误信息
实现错误重试机制
避免记录敏感信息
使用结构化日志
设置适当的日志级别
Files:
src/main/presenter/configPresenter/modelDefaultSettings.ts
src/main/**/*.ts
📄 CodeRabbit Inference Engine (CLAUDE.md)
Main to Renderer: Use EventBus to broadcast events via mainWindow.webContents.send()
Use Electron's built-in APIs for file system and native dialogs
Files:
src/main/presenter/configPresenter/modelDefaultSettings.ts
src/main/presenter/**/*.ts
📄 CodeRabbit Inference Engine (CLAUDE.md)
One presenter per functional domain
Files:
src/main/presenter/configPresenter/modelDefaultSettings.ts
src/main/presenter/configPresenter/**/*.ts
📄 CodeRabbit Inference Engine (CLAUDE.md)
Centralize configuration in configPresenter/
Files:
src/main/presenter/configPresenter/modelDefaultSettings.ts
**/*.{js,jsx,ts,tsx}
📄 CodeRabbit Inference Engine (.cursor/rules/development-setup.mdc)
**/*.{js,jsx,ts,tsx}: 使用 OxLint 进行代码检查
Log和注释使用英文书写
Files:
src/main/presenter/configPresenter/modelDefaultSettings.ts
src/{main,renderer}/**/*.ts
📄 CodeRabbit Inference Engine (.cursor/rules/electron-best-practices.mdc)
src/{main,renderer}/**/*.ts: Use context isolation for improved security
Implement proper inter-process communication (IPC) patterns
Optimize application startup time with lazy loading
Implement proper error handling and logging for debugging
Files:
src/main/presenter/configPresenter/modelDefaultSettings.ts
src/main/**/*.{ts,js,tsx,jsx}
📄 CodeRabbit Inference Engine (.cursor/rules/project-structure.mdc)
主进程代码放在
src/main
Files:
src/main/presenter/configPresenter/modelDefaultSettings.ts
🧠 Learnings (1)
src/main/presenter/configPresenter/modelDefaultSettings.ts (1)
Learnt from: CR
PR: ThinkInAIXYZ/deepchat#0
File: CLAUDE.md:0-0
Timestamp: 2025-07-21T01:45:33.790Z
Learning: Applies to src/main/presenter/configPresenter/**/*.ts : Centralize configuration in configPresenter/
🔇 Additional comments (1)
src/main/presenter/configPresenter/modelDefaultSettings.ts (1)
355-357: Same off-by-one risk for Gemini 2.5 FlashFor consistency with the comment above, verify that 65 535 is indeed the hard limit for
models/gemini-2.5-flash. If the API still allows 65 536, user prompts that expect the full range will be silently clipped.
| maxTokens: 65535, | ||
| contextLength: 1048576, | ||
| match: ['gemini-2.5-pro'], |
There was a problem hiding this comment.
💡 Verification agent
🧩 Analysis chain
Double-check the new 65 535 limit – documentation appears inconsistent
The previous value (65 536) was reduced by one. Google’s Vertex AI docs for Gemini 2.5 Pro list both 65 535 and 65 536 in different places, depending on whether they count the EOS token. Please confirm the SDK you call actually enforces 65 535, otherwise this change may re-introduce truncation or a 400-error on larger responses.
🏁 Script executed:
#!/bin/bash
# Scan codebase for runtime overrides that still assume 65536
rg -n --pretty $'65536' | headLength of output: 2888
Inconsistent maxTokens values – confirm and unify
We updated the default in modelDefaultSettings.ts to 65 535, but geminiProvider.ts still uses 65 536. Please verify which limit the Vertex AI SDK actually enforces (with or without EOS) and align both places to avoid unexpected truncation or 400-errors.
• src/main/presenter/configPresenter/modelDefaultSettings.ts (lines 343–345)
maxTokens: 65535,
contextLength: 1048576,
match: ['gemini-2.5-pro'],• src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts (lines 61, 73)
maxTokens: 65536,Next steps:
- Confirm via the Vertex AI SDK or docs whether the true max is 65 535 or 65 536 (EOS-inclusive vs exclusive).
- Update both the default settings and any hardcoded overrides to the correct value.
🤖 Prompt for AI Agents
In src/main/presenter/configPresenter/modelDefaultSettings.ts lines 343-345 and
src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts lines 61 and
73, the maxTokens values are inconsistent (65535 vs 65536). Verify the exact
maxTokens limit enforced by the Vertex AI SDK, considering whether it includes
the EOS token or not. Then update both files to use the confirmed correct
maxTokens value to ensure consistency and prevent truncation or errors.
2 participants
在使用Gemini Flash 和Pro访问的时候,有概率触发下面的异常
根据Google 的API 文档里面 flash 和Pro的 max output token 会少一个
https://cloud.google.com/vertex-ai/generative-ai/docs/models/gemini/2-5-flash
同时修复gemini-2.5-flash-lite-preview-06-17 的值
Summary by CodeRabbit