feat: support gemini upstream Provider DB #984
Conversation
WalkthroughProvider DB is pruned and standardized (modalities, limits, costs, support flags). Config layer now resolves provider aliases and model IDs (including Google/Gemini normalization) and exposes capability checks. Renderer model type reflects image-output capability. Gemini provider refactors to use DB-backed model mapping, normalized IDs, capability-driven config, and updated response parsing. UI tweaks thinkingBudget validation for Gemini. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
actor User
participant UI as Settings UI
participant CP as Config Presenter
participant MC as ModelCapabilities
participant DB as Provider DB
participant GP as Gemini Provider
participant Google as Google API
User->>UI: Select provider/model
UI->>CP: getDbProviderModels(providerId)
CP->>MC: resolveProviderId(providerId)
MC-->>CP: normalizedProviderId
CP->>DB: read providers[normalizedProviderId]
DB-->>CP: provider models
CP->>MC: supportsImageOutput(providerId, modelId)?
MC-->>CP: true/false
CP-->>UI: models with type Chat/ImageGeneration
User->>GP: Generate (modelId)
GP->>GP: ensureGoogleModelName(models/... prefix)
GP->>MC: supportsToolCall/vision/imageOutput(model)
MC-->>GP: capability flags
alt No models from API
GP->>DB: fallback to provider DB models
end
GP->>Google: generateContent(stream?) with normalized model + config
Google-->>GP: candidates/stream (incl. thinking/content)
GP->>GP: parse thinking (legacy/new), tools results
GP-->>User: response/content (images/text/errors)
Estimated code review effort🎯 4 (Complex) | ⏱️ ~60 minutes Possibly related PRs
Suggested reviewers
Poem
Pre-merge checks and finishing touches✅ Passed checks (3 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
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. Comment |
![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 (2)
src/renderer/src/components/settings/ModelConfigDialog.vue (2)
597-597: Consider using capability-based checks instead of hard-coded provider ID.The hard-coded check
props.providerId?.toLowerCase() === 'gemini'may not align with provider alias resolution (e.g.,gemini→modelConfig.ts). If the resolved provider ID isgemini, this check would fail.Consider querying a capability flag or checking the resolved provider ID:
-const isGeminiProvider = computed(() => props.providerId?.toLowerCase() === 'gemini') +const isGeminiProvider = computed(() => { + const resolvedId = props.providerId?.toLowerCase() + return resolvedId === 'gemini' || resolvedId === 'google' +})Or better yet, expose a capability check from
configPresentersuch assupportsUnboundedThinkingBudget(providerId, modelId)for clearer semantics.
671-673: Add clarifying comment for Gemini’s unbounded thinkingBudgetIn
ModelConfigDialog.vue, annotate the-1special case:if (!range) return '' +// Gemini supports –1 as an unbounded (dynamic) thinking budget if (isGeminiProvider.value && value === -1) { return '' }
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (6)
resources/model-db/providers.json(18 hunks)src/main/presenter/configPresenter/index.ts(2 hunks)src/main/presenter/configPresenter/modelCapabilities.ts(3 hunks)src/main/presenter/configPresenter/modelConfig.ts(2 hunks)src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts(15 hunks)src/renderer/src/components/settings/ModelConfigDialog.vue(2 hunks)
🧰 Additional context used
📓 Path-based instructions (24)
src/renderer/src/**/*
📄 CodeRabbit inference engine (.cursor/rules/i18n.mdc)
src/renderer/src/**/*: All user-facing strings must use i18n keys (avoid hardcoded user-visible text in code)
Use the 'vue-i18n' framework for all internationalization in the renderer
Ensure all user-visible text in the renderer uses the translation system
Files:
src/renderer/src/components/settings/ModelConfigDialog.vue
src/renderer/**/*.{vue,ts,js,tsx,jsx}
📄 CodeRabbit inference engine (.cursor/rules/project-structure.mdc)
渲染进程代码放在
src/renderer
Files:
src/renderer/src/components/settings/ModelConfigDialog.vue
src/renderer/src/**/*.{vue,ts,tsx,js,jsx}
📄 CodeRabbit inference engine (.cursor/rules/vue-best-practices.mdc)
src/renderer/src/**/*.{vue,ts,tsx,js,jsx}: Use the Composition API for better code organization and reusability
Implement proper state management with Pinia
Utilize Vue Router for navigation and route management
Leverage Vue's built-in reactivity system for efficient data handling
Files:
src/renderer/src/components/settings/ModelConfigDialog.vue
src/renderer/src/**/*.vue
📄 CodeRabbit inference engine (.cursor/rules/vue-best-practices.mdc)
Use scoped styles to prevent CSS conflicts between components
Files:
src/renderer/src/components/settings/ModelConfigDialog.vue
src/renderer/**/*.{ts,tsx,vue}
📄 CodeRabbit inference engine (.cursor/rules/vue-shadcn.mdc)
src/renderer/**/*.{ts,tsx,vue}: Use descriptive variable names with auxiliary verbs (e.g., isLoading, hasError).
Use TypeScript for all code; prefer types over interfaces.
Avoid enums; use const objects instead.
Use arrow functions for methods and computed properties.
Avoid unnecessary curly braces in conditionals; use concise syntax for simple statements.
Files:
src/renderer/src/components/settings/ModelConfigDialog.vue
src/renderer/**/*.{vue,ts}
📄 CodeRabbit inference engine (.cursor/rules/vue-shadcn.mdc)
Implement lazy loading for routes and components.
Files:
src/renderer/src/components/settings/ModelConfigDialog.vue
src/renderer/**/*.{ts,vue}
📄 CodeRabbit inference engine (.cursor/rules/vue-shadcn.mdc)
src/renderer/**/*.{ts,vue}: Use useFetch and useAsyncData for data fetching.
Implement SEO best practices using Nuxt's useHead and useSeoMeta.Use Pinia for frontend state management (do not introduce alternative state libraries)
Files:
src/renderer/src/components/settings/ModelConfigDialog.vue
**/*.{ts,tsx,js,vue}
📄 CodeRabbit inference engine (CLAUDE.md)
Use English for all logs and comments
Files:
src/renderer/src/components/settings/ModelConfigDialog.vuesrc/main/presenter/configPresenter/modelConfig.tssrc/main/presenter/configPresenter/index.tssrc/main/presenter/configPresenter/modelCapabilities.tssrc/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
**/*.{ts,tsx,vue}
📄 CodeRabbit inference engine (CLAUDE.md)
Enable and adhere to strict TypeScript typing (avoid implicit any, prefer precise types)
Files:
src/renderer/src/components/settings/ModelConfigDialog.vuesrc/main/presenter/configPresenter/modelConfig.tssrc/main/presenter/configPresenter/index.tssrc/main/presenter/configPresenter/modelCapabilities.tssrc/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
src/renderer/{src,shell,floating}/**/*.vue
📄 CodeRabbit inference engine (CLAUDE.md)
src/renderer/{src,shell,floating}/**/*.vue: Use Vue 3 Composition API for all components
All user-facing strings must use i18n keys via vue-i18n (no hard-coded UI strings)
Use Tailwind CSS utilities and ensure styles are scoped in Vue components
Files:
src/renderer/src/components/settings/ModelConfigDialog.vue
src/renderer/src/components/**/*
📄 CodeRabbit inference engine (CLAUDE.md)
Organize UI components by feature within src/renderer/src/
Files:
src/renderer/src/components/settings/ModelConfigDialog.vue
src/renderer/src/**
📄 CodeRabbit inference engine (AGENTS.md)
Put application code for the Vue app under src/renderer/src (components, stores, views, i18n, lib)
Files:
src/renderer/src/components/settings/ModelConfigDialog.vue
src/renderer/src/**/*.{vue,ts}
📄 CodeRabbit inference engine (AGENTS.md)
All user-facing strings in the renderer must use vue-i18n keys defined in src/renderer/src/i18n
Files:
src/renderer/src/components/settings/ModelConfigDialog.vue
**/*.{js,jsx,ts,tsx,vue}
📄 CodeRabbit inference engine (AGENTS.md)
Apply Prettier formatting: single quotes, no semicolons, max width 100
Files:
src/renderer/src/components/settings/ModelConfigDialog.vuesrc/main/presenter/configPresenter/modelConfig.tssrc/main/presenter/configPresenter/index.tssrc/main/presenter/configPresenter/modelCapabilities.tssrc/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
src/renderer/**/*.vue
📄 CodeRabbit inference engine (AGENTS.md)
Name Vue components in PascalCase (e.g., ChatInput.vue)
Files:
src/renderer/src/components/settings/ModelConfigDialog.vue
**/*.{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
**/*.{js,jsx,ts,tsx}: Use OxLint for JS/TS code; pre-commit hooks run lint-staged and typecheck
Use camelCase for variables and functions
Use PascalCase for types and classes
Use SCREAMING_SNAKE_CASE for constants
Files:
src/main/presenter/configPresenter/modelConfig.tssrc/main/presenter/configPresenter/index.tssrc/main/presenter/configPresenter/modelCapabilities.tssrc/main/presenter/llmProviderPresenter/providers/geminiProvider.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/modelConfig.tssrc/main/presenter/configPresenter/index.tssrc/main/presenter/configPresenter/modelCapabilities.tssrc/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
src/main/**/*.ts
📄 CodeRabbit inference engine (.cursor/rules/electron-best-practices.mdc)
Use Electron's built-in APIs for file system and native dialogs
Files:
src/main/presenter/configPresenter/modelConfig.tssrc/main/presenter/configPresenter/index.tssrc/main/presenter/configPresenter/modelCapabilities.tssrc/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
**/*.{ts,tsx}
📄 CodeRabbit inference engine (.cursor/rules/error-logging.mdc)
**/*.{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/modelConfig.tssrc/main/presenter/configPresenter/index.tssrc/main/presenter/configPresenter/modelCapabilities.tssrc/main/presenter/llmProviderPresenter/providers/geminiProvider.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/modelConfig.tssrc/main/presenter/configPresenter/index.tssrc/main/presenter/configPresenter/modelCapabilities.tssrc/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
src/main/**
📄 CodeRabbit inference engine (AGENTS.md)
Place all Electron main-process code under src/main/
Files:
src/main/presenter/configPresenter/modelConfig.tssrc/main/presenter/configPresenter/index.tssrc/main/presenter/configPresenter/modelCapabilities.tssrc/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
src/main/presenter/**
📄 CodeRabbit inference engine (AGENTS.md)
src/main/presenter/**: Organize main-process presenters under src/main/presenter/ (Window/Tab/Thread/Mcp/Config/LLMProvider)
Follow the Presenter pattern for main-process modules
Files:
src/main/presenter/configPresenter/modelConfig.tssrc/main/presenter/configPresenter/index.tssrc/main/presenter/configPresenter/modelCapabilities.tssrc/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
{build,resources,out,dist}/**
📄 CodeRabbit inference engine (AGENTS.md)
Keep build outputs and assets within build/, resources/, out/, and dist/
Files:
resources/model-db/providers.json
src/main/presenter/llmProviderPresenter/providers/*.ts
📄 CodeRabbit inference engine (.cursor/rules/llm-agent-loop.mdc)
src/main/presenter/llmProviderPresenter/providers/*.ts: Each file insrc/main/presenter/llmProviderPresenter/providers/*.tsshould handle interaction with a specific LLM API, including request/response formatting, tool definition conversion, native/non-native tool call management, and standardizing output streams to a common event format.
Provider implementations must use acoreStreammethod that yields standardized stream events to decouple the main loop from provider-specific details.
ThecoreStreammethod in each Provider must perform a single streaming API request per conversation round and must not contain multi-round tool call loop logic.
Provider files should implement helper methods such asformatMessages,convertToProviderTools,parseFunctionCalls, andprepareFunctionCallPromptas needed for provider-specific logic.
All provider implementations must parse provider-specific data chunks and yield standardized events for text, reasoning, tool calls, usage, errors, stop reasons, and image data.
When a provider does not support native function calling, it must prepare messages using prompt wrapping (e.g.,prepareFunctionCallPrompt) before making the API call.
When a provider supports native function calling, MCP tools must be converted to the provider's format (e.g., usingconvertToProviderTools) and included in the API request.
Provider implementations should aggregate and yield usage events as part of the standardized stream.
Provider implementations should yield image data events in the standardized format when applicable.
Provider implementations should yield reasoning events in the standardized format when applicable.
Provider implementations should yield tool call events (tool_call_start,tool_call_chunk,tool_call_end) in the standardized format.
Provider implementations should yield stop events with appropriatestop_reasonin the standardized format.
Provider implementations should yield error events in the standardized format...
Files:
src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
🧠 Learnings (8)
📚 Learning: 2025-07-21T01:46:52.880Z
Learnt from: CR
PR: ThinkInAIXYZ/deepchat#0
File: .cursor/rules/llm-agent-loop.mdc:0-0
Timestamp: 2025-07-21T01:46:52.880Z
Learning: Applies to src/main/presenter/llmProviderPresenter/streamEvents.ts : Standardized stream events should conform to the `LLMCoreStreamEvent` interface, ideally defined in a shared file such as `src/main/presenter/llmProviderPresenter/streamEvents.ts`.
Applied to files:
src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
📚 Learning: 2025-09-06T03:07:23.817Z
Learnt from: CR
PR: ThinkInAIXYZ/deepchat#0
File: CLAUDE.md:0-0
Timestamp: 2025-09-06T03:07:23.817Z
Learning: Applies to src/main/presenter/llmProviderPresenter/providers/*.ts : Each LLM provider implementation must expose a coreStream method following the standardized event interface
Applied to files:
src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
📚 Learning: 2025-07-21T01:46:52.880Z
Learnt from: CR
PR: ThinkInAIXYZ/deepchat#0
File: .cursor/rules/llm-agent-loop.mdc:0-0
Timestamp: 2025-07-21T01:46:52.880Z
Learning: Applies to src/main/presenter/llmProviderPresenter/providers/*.ts : Provider implementations must use a `coreStream` method that yields standardized stream events to decouple the main loop from provider-specific details.
Applied to files:
src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
📚 Learning: 2025-07-21T01:46:52.880Z
Learnt from: CR
PR: ThinkInAIXYZ/deepchat#0
File: .cursor/rules/llm-agent-loop.mdc:0-0
Timestamp: 2025-07-21T01:46:52.880Z
Learning: Applies to src/main/presenter/llmProviderPresenter/index.ts : `src/main/presenter/llmProviderPresenter/index.ts` should manage the overall Agent loop, conversation history, tool execution via `McpPresenter`, and frontend communication via `eventBus`.
Applied to files:
src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
📚 Learning: 2025-07-21T01:46:52.880Z
Learnt from: CR
PR: ThinkInAIXYZ/deepchat#0
File: .cursor/rules/llm-agent-loop.mdc:0-0
Timestamp: 2025-07-21T01:46:52.880Z
Learning: Applies to src/main/presenter/llmProviderPresenter/index.ts : The main Agent loop should send standardized `STREAM_EVENTS` (`RESPONSE`, `END`, `ERROR`) to the frontend via `eventBus`.
Applied to files:
src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
📚 Learning: 2025-07-21T01:46:52.880Z
Learnt from: CR
PR: ThinkInAIXYZ/deepchat#0
File: .cursor/rules/llm-agent-loop.mdc:0-0
Timestamp: 2025-07-21T01:46:52.880Z
Learning: Applies to src/main/presenter/llmProviderPresenter/providers/*.ts : Provider implementations should aggregate and yield usage events as part of the standardized stream.
Applied to files:
src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
📚 Learning: 2025-09-04T11:03:30.184Z
Learnt from: CR
PR: ThinkInAIXYZ/deepchat#0
File: .cursor/rules/provider-guidelines.mdc:0-0
Timestamp: 2025-09-04T11:03:30.184Z
Learning: Integrate via the llmProviderPresenter entry point (src/main/presenter/llmProviderPresenter/index.ts) as the related implementation entry
Applied to files:
src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
📚 Learning: 2025-07-21T01:46:52.880Z
Learnt from: CR
PR: ThinkInAIXYZ/deepchat#0
File: .cursor/rules/llm-agent-loop.mdc:0-0
Timestamp: 2025-07-21T01:46:52.880Z
Learning: Applies to src/main/presenter/llmProviderPresenter/providers/*.ts : Each file in `src/main/presenter/llmProviderPresenter/providers/*.ts` should handle interaction with a specific LLM API, including request/response formatting, tool definition conversion, native/non-native tool call management, and standardizing output streams to a common event format.
Applied to files:
src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
🧬 Code graph analysis (2)
src/main/presenter/configPresenter/index.ts (1)
src/main/presenter/configPresenter/modelCapabilities.ts (1)
modelCapabilities(123-123)
src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts (2)
src/shared/types/presenters/llmprovider.presenter.d.ts (1)
MODEL_META(27-42)src/main/presenter/configPresenter/modelCapabilities.ts (1)
modelCapabilities(123-123)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: build-check (x64)
🔇 Additional comments (12)
src/main/presenter/configPresenter/index.ts (2)
665-667: LGTM! Provider ID resolution correctly uses alias mapping.The code properly resolves provider aliases (e.g.,
gemini→modelCapabilities.resolveProviderIdwith a safe fallback to the lowercased original ID. This ensures consistent provider lookup in the DB.
683-686: LGTM! Capability-driven model type determination.The conditional type assignment based on
modalities.outputcorrectly identifies image generation models. This aligns with the broader PR goal of standardizing model capabilities via the Provider DB.src/main/presenter/configPresenter/modelConfig.ts (2)
24-25: LGTM! Gemini provider alias is consistent across the codebase.The
gemini: 'google'alias addition matches the pattern inmodelCapabilities.ts(lines 21-22) and supports unified provider resolution throughout the application.
242-244: LGTM! Model ID normalization handles Google Gemini SDK format.The code correctly strips the
models/prefix that Google Gemini SDK returns, enabling consistent model ID lookups. The inline comment clearly explains the purpose.src/main/presenter/configPresenter/modelCapabilities.ts (2)
61-65: LGTM! Making resolveProviderId public enables centralized alias resolution.The method is now used by
configPresenter/index.ts(line 666) to resolve provider aliases when accessing the Provider DB. This centralizes alias handling logic and prevents duplication.
103-120: LGTM! New capability methods follow consistent patterns.The three new methods (
supportsVision,supportsToolCall,supportsImageOutput) enable capability-driven model configuration. They follow the same pattern as existing methods and correctly query the Provider DB model definitions.src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts (6)
70-72: LGTM! Centralized model name normalization for Google API.The
ensureGoogleModelNamehelper consistently prefixes model IDs withmodels/as required by the Google Gemini SDK. This prevents errors from missing prefixes and centralizes the normalization logic.
88-101: LGTM! Provider DB fallback ensures robustness.The fallback to
getDbProviderModelswhen the Gemini API returns no models is a good defensive measure. The mapping correctly populatesMODEL_METAfields from the Provider DB configuration, maintaining functionality even when the API is unavailable.
105-148: LGTM! Capability-driven model mapping improves maintainability.The refactored model mapping queries
modelCapabilitiesfor each capability (vision, tool call, reasoning, image output) rather than hard-coding assumptions. This makes the code data-driven and easier to maintain as the Provider DB evolves.Minor observation: The group categorization logic (lines 132-133) uses regex patterns that could be fragile if Gemini changes naming conventions. Consider whether group information should also come from the Provider DB.
312-318: LGTM! Thinking budget support check is capability-driven.The
supportsThinkingBudgetmethod correctly queriesmodelCapabilities.getThinkingBudgetRangeto determine support, replacing hard-coded model name checks. This aligns with the broader refactor toward capability-driven configuration.
848-858: LGTM! Conditional function calling configuration prevents empty tool config.The guard clause checking
hasFunctionDeclarationsbefore configuringfunctionCallingConfigis a good defensive measure. It prevents sending tool configuration to the Gemini API when there are no actual function declarations, avoiding potential API errors or unexpected behavior.
250-274: Verify the intentional removal of automatic model enablement.The updated
autoEnableModelsIfNeededmethod no longer automatically enables models for Gemini, requiring users to manually enable them. This is a breaking behavioral change that could impact user experience, especially for new installations.Please confirm:
- This behavioral change is intentional and aligns with product requirements
- There is user-facing documentation or UI guidance explaining that users need to manually enable Gemini models
- The console log message (line 272) provides sufficient guidance
Consider whether there should be a notification or toast in the UI directing users to enable models, rather than just a console message they may not see.
| "default": true, | ||
| "budget": { | ||
| "default": -1, | ||
| "min": 0, | ||
| "max": 24576 | ||
| } | ||
| }, | ||
| "search": { | ||
| "supported": true, | ||
| "default": false |
There was a problem hiding this comment.
Fix reasoning budget defaults to satisfy validation bounds
reasoning.budget.default is -1 while min is 0, so the default is outside the declared range. The UI layer now validates thinkingBudget against these bounds, and this mismatch causes the Gemini config to be rejected/flagged immediately. The same issue appears in the other Gemini 2.5 entries where default is -1 and min is 0/128/512. Please align the defaults with the declared minimums (or relax the minimum) before landing. Example fix:
- "default": -1,
- "min": 0,
+ "default": 0,
+ "min": 0,
"max": 24576(Be sure to adjust the corresponding budget blocks for the other Gemini 2.5 variants as well.)
📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| "default": true, | |
| "budget": { | |
| "default": -1, | |
| "min": 0, | |
| "max": 24576 | |
| } | |
| }, | |
| "search": { | |
| "supported": true, | |
| "default": false | |
| "default": true, | |
| "budget": { | |
| "default": 0, | |
| "min": 0, | |
| "max": 24576 | |
| } | |
| }, | |
| "search": { | |
| "supported": true, | |
| "default": false |
🤖 Prompt for AI Agents
In resources/model-db/providers.json around lines 12594 to 12603, the
reasoning.budget.default is set to -1 while the declared min is 0 (making the
default invalid); update reasoning.budget.default to be at least the declared
min (e.g., set to 0 or the appropriate min for Gemini 2.5 variants) and apply
the same change to the other Gemini 2.5 entries that currently have default -1
so all reasoning.budget.default values fall within their corresponding min..max
ranges.
|
@codex review |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
deepchat/src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts
Lines 343 to 352 in 667424b
Skip sending negative thinking budgets to Gemini API
The provider DB now ships Gemini models with a default thinking budget of -1 to signal “use the server default”, and the settings dialog allows users to keep -1 without validation errors. However getGenerateContentConfig still forwards any numeric value (including -1) in thinkingBudget whenever the model supports reasoning. The Gemini API expects a non‑negative token budget and will reject calls that include a negative number, so reasoning‑capable models will start failing as soon as the new defaults are loaded. Consider omitting the thinkingBudget field whenever the value is less than 0 instead of passing it through.
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting
support gemini upstream Provider DB
Summary by CodeRabbit