Skip to content

feat: add thinkingLevel and thinkingBudget config for Gemini models #1110

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
brendan-kellam merged 5 commits into from
Apr 13, 2026
Merged

feat: add thinkingLevel and thinkingBudget config for Gemini models #1110

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
ffdbbed
feat: add thinkingLevel and thinkingBudget config for Gemini models
brendan-kellam Apr 13, 2026
1888acc
docs: add changelog entry for thinkingLevel/thinkingBudget
brendan-kellam Apr 13, 2026
46cf17a
remove GOOGLE_VERTEX_INCLUDE_THOUGHTS
brendan-kellam Apr 13, 2026
3efe2ac
docs: add changelog entry for GOOGLE_VERTEX_INCLUDE_THOUGHTS removal
brendan-kellam Apr 13, 2026
8204819
docs
brendan-kellam Apr 13, 2026
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
7 changes: 7 additions & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,13 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0

## [Unreleased]

### Added
- Added `thinkingLevel` and `thinkingBudget` configuration options for Google Generative AI and Google Vertex providers. [#1110](https://github.com/sourcebot-dev/sourcebot/pull/1110)

### Changed
- Deprecated `GOOGLE_VERTEX_THINKING_BUDGET_TOKENS` environment variable in favor of per-model `thinkingBudget` config. [#1110](https://github.com/sourcebot-dev/sourcebot/pull/1110)
- Removed `GOOGLE_VERTEX_INCLUDE_THOUGHTS` environment variable. Thoughts are now always included. [#1110](https://github.com/sourcebot-dev/sourcebot/pull/1110)

## [4.16.8] - 2026-04-09

### Added
Expand Down
16 changes: 14 additions & 2 deletions docs/docs/configuration/language-model-providers.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -157,6 +157,10 @@ The `reasoningSummary` field controls whether the model returns its reasoning pr

[Vercel AI SDK Google Generative AI Docs](https://ai-sdk.dev/providers/ai-sdk-providers/google-generative-ai)

The `thinkingLevel` field controls the depth of reasoning for Gemini 3 models. Valid values are `minimal`, `low`, `medium`, and `high`. See [thinking levels](https://ai.google.dev/gemini-api/docs/thinking#thinking-levels).

The `thinkingBudget` field sets the number of thinking tokens for Gemini 2.5 models. Set to `-1` for dynamic thinking. See [thinking budget](https://ai.google.dev/gemini-api/docs/thinking#set-budget).

```json wrap icon="code" Example config with Google Generative AI provider
{
"$schema": "https://raw.githubusercontent.com/sourcebot-dev/sourcebot/main/schemas/v3/index.json",
Expand All @@ -168,7 +172,9 @@ The `reasoningSummary` field controls whether the model returns its reasoning pr
"token": {
"env": "GOOGLE_GENERATIVE_AI_API_KEY"
},
"baseUrl": "OPTIONAL_BASE_URL"
"baseUrl": "OPTIONAL_BASE_URL",
"thinkingLevel": "high", // "minimal" | "low" | "medium" | "high" (Gemini 3 models)
"thinkingBudget": -1 // number of thinking tokens, -1 for dynamic (Gemini 2.5 models)
}
]
}
Expand All @@ -181,6 +187,10 @@ The `reasoningSummary` field controls whether the model returns its reasoning pr

[Vercel AI SDK Google Vertex AI Docs](https://ai-sdk.dev/providers/ai-sdk-providers/google-vertex)

The `thinkingLevel` field controls the depth of reasoning for Gemini 3 models. Valid values are `minimal`, `low`, `medium`, and `high`. See [thinking levels](https://ai.google.dev/gemini-api/docs/thinking#thinking-levels).

The `thinkingBudget` field sets the number of thinking tokens for Gemini 2.5 models. Set to `-1` for dynamic thinking. See [thinking budget](https://ai.google.dev/gemini-api/docs/thinking#set-budget).

```json wrap icon="code" Example config with Google Vertex provider
{
"$schema": "https://raw.githubusercontent.com/sourcebot-dev/sourcebot/main/schemas/v3/index.json",
Expand All @@ -194,7 +204,9 @@ The `reasoningSummary` field controls whether the model returns its reasoning pr
"credentials": {
"env": "GOOGLE_APPLICATION_CREDENTIALS"
},
"baseUrl": "OPTIONAL_BASE_URL"
"baseUrl": "OPTIONAL_BASE_URL",
"thinkingLevel": "high", // "minimal" | "low" | "medium" | "high" (Gemini 3 models)
"thinkingBudget": -1 // number of thinking tokens, -1 for dynamic (Gemini 2.5 models)
}
]
}
Expand Down
56 changes: 56 additions & 0 deletions docs/snippets/schemas/v3/index.schema.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2300,6 +2300,20 @@
"pattern": "^https?:\\/\\/[^\\s/$.?#].[^\\s]*$",
"description": "Optional base URL."
},
"thinkingLevel": {
"type": "string",
"description": "Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels",
"enum": [
"minimal",
"low",
"medium",
"high"
]
},
"thinkingBudget": {
"type": "integer",
"description": "Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget"
},
"temperature": {
"type": "number",
"description": "Optional temperature setting to use with the model."
Expand Down Expand Up @@ -2548,6 +2562,20 @@
"pattern": "^https?:\\/\\/[^\\s/$.?#].[^\\s]*$",
"description": "Optional base URL."
},
"thinkingLevel": {
"type": "string",
"description": "Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels",
"enum": [
"minimal",
"low",
"medium",
"high"
]
},
"thinkingBudget": {
"type": "integer",
"description": "Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget"
},
"temperature": {
"type": "number",
"description": "Optional temperature setting to use with the model."
Expand Down Expand Up @@ -3838,6 +3866,20 @@
"pattern": "^https?:\\/\\/[^\\s/$.?#].[^\\s]*$",
"description": "Optional base URL."
},
"thinkingLevel": {
"type": "string",
"description": "Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels",
"enum": [
"minimal",
"low",
"medium",
"high"
]
},
"thinkingBudget": {
"type": "integer",
"description": "Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget"
},
"temperature": {
"type": "number",
"description": "Optional temperature setting to use with the model."
Expand Down Expand Up @@ -4086,6 +4128,20 @@
"pattern": "^https?:\\/\\/[^\\s/$.?#].[^\\s]*$",
"description": "Optional base URL."
},
"thinkingLevel": {
"type": "string",
"description": "Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels",
"enum": [
"minimal",
"low",
"medium",
"high"
]
},
"thinkingBudget": {
"type": "integer",
"description": "Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget"
},
"temperature": {
"type": "number",
"description": "Optional temperature setting to use with the model."
Expand Down
56 changes: 56 additions & 0 deletions docs/snippets/schemas/v3/languageModel.schema.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -614,6 +614,20 @@
"pattern": "^https?:\\/\\/[^\\s/$.?#].[^\\s]*$",
"description": "Optional base URL."
},
"thinkingLevel": {
"type": "string",
"description": "Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels",
"enum": [
"minimal",
"low",
"medium",
"high"
]
},
"thinkingBudget": {
"type": "integer",
"description": "Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget"
},
"temperature": {
"type": "number",
"description": "Optional temperature setting to use with the model."
Expand Down Expand Up @@ -862,6 +876,20 @@
"pattern": "^https?:\\/\\/[^\\s/$.?#].[^\\s]*$",
"description": "Optional base URL."
},
"thinkingLevel": {
"type": "string",
"description": "Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels",
"enum": [
"minimal",
"low",
"medium",
"high"
]
},
"thinkingBudget": {
"type": "integer",
"description": "Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget"
},
"temperature": {
"type": "number",
"description": "Optional temperature setting to use with the model."
Expand Down Expand Up @@ -2152,6 +2180,20 @@
"pattern": "^https?:\\/\\/[^\\s/$.?#].[^\\s]*$",
"description": "Optional base URL."
},
"thinkingLevel": {
"type": "string",
"description": "Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels",
"enum": [
"minimal",
"low",
"medium",
"high"
]
},
"thinkingBudget": {
"type": "integer",
"description": "Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget"
},
"temperature": {
"type": "number",
"description": "Optional temperature setting to use with the model."
Expand Down Expand Up @@ -2400,6 +2442,20 @@
"pattern": "^https?:\\/\\/[^\\s/$.?#].[^\\s]*$",
"description": "Optional base URL."
},
"thinkingLevel": {
"type": "string",
"description": "Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels",
"enum": [
"minimal",
"low",
"medium",
"high"
]
},
"thinkingBudget": {
"type": "integer",
"description": "Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget"
},
"temperature": {
"type": "number",
"description": "Optional temperature setting to use with the model."
Expand Down
56 changes: 56 additions & 0 deletions packages/schemas/src/v3/index.schema.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2299,6 +2299,20 @@ const schema = {
"pattern": "^https?:\\/\\/[^\\s/$.?#].[^\\s]*$",
"description": "Optional base URL."
},
"thinkingLevel": {
"type": "string",
"description": "Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels",
"enum": [
"minimal",
"low",
"medium",
"high"
]
},
"thinkingBudget": {
"type": "integer",
"description": "Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget"
},
"temperature": {
"type": "number",
"description": "Optional temperature setting to use with the model."
Expand Down Expand Up @@ -2547,6 +2561,20 @@ const schema = {
"pattern": "^https?:\\/\\/[^\\s/$.?#].[^\\s]*$",
"description": "Optional base URL."
},
"thinkingLevel": {
"type": "string",
"description": "Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels",
"enum": [
"minimal",
"low",
"medium",
"high"
]
},
"thinkingBudget": {
"type": "integer",
"description": "Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget"
},
"temperature": {
"type": "number",
"description": "Optional temperature setting to use with the model."
Expand Down Expand Up @@ -3837,6 +3865,20 @@ const schema = {
"pattern": "^https?:\\/\\/[^\\s/$.?#].[^\\s]*$",
"description": "Optional base URL."
},
"thinkingLevel": {
"type": "string",
"description": "Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels",
"enum": [
"minimal",
"low",
"medium",
"high"
]
},
"thinkingBudget": {
"type": "integer",
"description": "Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget"
},
"temperature": {
"type": "number",
"description": "Optional temperature setting to use with the model."
Expand Down Expand Up @@ -4085,6 +4127,20 @@ const schema = {
"pattern": "^https?:\\/\\/[^\\s/$.?#].[^\\s]*$",
"description": "Optional base URL."
},
"thinkingLevel": {
"type": "string",
"description": "Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels",
"enum": [
"minimal",
"low",
"medium",
"high"
]
},
"thinkingBudget": {
"type": "integer",
"description": "Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget"
},
"temperature": {
"type": "number",
"description": "Optional temperature setting to use with the model."
Expand Down
16 changes: 16 additions & 0 deletions packages/schemas/src/v3/index.type.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -962,6 +962,14 @@ export interface GoogleGenerativeAILanguageModel {
* Optional base URL.
*/
baseUrl?: string;
/**
* Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels
*/
thinkingLevel?: "minimal" | "low" | "medium" | "high";
/**
* Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget
*/
thinkingBudget?: number;
/**
* Optional temperature setting to use with the model.
*/
Expand Down Expand Up @@ -1056,6 +1064,14 @@ export interface GoogleVertexLanguageModel {
* Optional base URL.
*/
baseUrl?: string;
/**
* Optional thinking level for Gemini 3 models. Controls the depth of reasoning the model performs. See https://ai.google.dev/gemini-api/docs/thinking#thinking-levels
*/
thinkingLevel?: "minimal" | "low" | "medium" | "high";
/**
* Optional thinking budget for Gemini 2.5 models. Sets the number of thinking tokens the model can use when generating a response. Set to -1 for dynamic thinking. See https://ai.google.dev/gemini-api/docs/thinking#set-budget
*/
thinkingBudget?: number;
/**
* Optional temperature setting to use with the model.
*/
Expand Down
Loading
Loading