feat: Update AI model support and configuration options across providers

gianged · gianged · commit fab299dc0eea · 2026-03-12T16:57:41.000+07:00
diff --git a/CLAUDE.md b/CLAUDE.md
@@ -6,12 +6,13 @@ This file provides guidance to Claude Code (claude.ai/code) when working with co
 
 **Universal Commit Assistant** is a VS Code extension that generates AI-powered commit messages using multiple AI providers. The extension supports 8 languages and follows modern TypeScript development practices with automated release workflows.
 
-## Latest Updates (November 2025)
+## Latest Updates (March 2026)
 
-- **Qwen Provider Added**: Integration with Alibaba Cloud DashScope API supporting Qwen Max, Qwen Plus, and Qwen Turbo models
-- **GPT-5.1 Models**: Updated OpenAI provider with GPT-5.1, GPT-5.1 Codex, and GPT-5.1 Codex Mini
-- **Gemini 3 Models**: Updated Google Gemini provider with Gemini 3 Pro (flagship) and preview models
-- **Enhanced Model Support**: Updated to support latest AI models from all major providers
+- **Gemini 3.1 Pro & 3 Flash**: Added latest Google Gemini models with 1M context window
+- **Claude Opus 4.6 & Sonnet 4.6**: Updated Anthropic provider with latest Claude models
+- **GPT-5.4**: Added OpenAI GPT-5.4 as new default model
+- **Deep Analysis Default**: Increased maxDiffLength to 100k chars for comprehensive code analysis
+- **Code Cleanup**: Removed dead code, fixed progress bug, added cloud provider timeouts, added maxTokens to Gemini/Ollama
 
 ## Development Commands
 
@@ -93,13 +94,13 @@ src/
 ├── providers/            # AI provider implementations
 │   ├── aiProviderFactory.ts   # Factory for creating provider instances
 │   ├── baseProvider.ts        # Abstract base class
-│   ├── anthropicProvider.ts   # Claude 4.5 Haiku/Sonnet/Opus
+│   ├── anthropicProvider.ts   # Claude Haiku 4.5/Sonnet 4.6/Opus 4.6
 │   ├── deepseekProvider.ts    # DeepSeek V3.1 models
-│   ├── geminiProvider.ts      # Google Gemini 3 models
+│   ├── geminiProvider.ts      # Google Gemini 3.1/3/2.5 models
 │   ├── lmstudioProvider.ts    # Local LM Studio integration
 │   ├── mistralProvider.ts     # Mistral AI models
 │   ├── ollamaProvider.ts      # Local Ollama integration
-│   ├── openaiProvider.ts      # OpenAI GPT-5.1 models
+│   ├── openaiProvider.ts      # OpenAI GPT-5.4/5.1 models
 │   ├── openrouterProvider.ts  # OpenRouter proxy service
 │   └── qwenProvider.ts        # Alibaba Qwen models
 ├── services/
@@ -152,26 +153,28 @@ Provider-specific settings use nested structure:
 - Cloud providers: `universal-commit-assistant.openai.model`, `universal-commit-assistant.anthropic.model`, `universal-commit-assistant.deepseek.model`, `universal-commit-assistant.qwen.model`
 - Local providers: `universal-commit-assistant.ollama.baseUrl`, `universal-commit-assistant.lmstudio.model`
 
-### Latest AI Model Updates (November 2025)
+### Latest AI Model Updates (March 2026)
 
 **OpenAI Latest Models:**
 
-- **GPT-5.1** (gpt-5.1) - Latest model balancing intelligence and speed, released November 2025
+- **GPT-5.4** (gpt-5.4) - Latest model, best intelligence and speed (default)
+- **GPT-5.1** (gpt-5.1) - Previous generation, excellent balance
 - **GPT-5.1 Codex** (gpt-5.1-codex) - Optimized for coding tasks
 - **GPT-5.1 Codex Mini** (gpt-5.1-codex-mini) - Fast coding model
 - **GPT-5 Mini** (gpt-5-mini) - Fast and cost-effective, 400K context window
-- **GPT-5** (gpt-5) - Full capability flagship model
+- **GPT-5** (gpt-5) - Previous flagship model
 
 **Anthropic Latest Models:**
 
-- **Claude Haiku 4.5** (claude-haiku-4-5-20251001) - Fast and cost-effective, released October 2025, 200K context window
-- **Claude Sonnet 4.5** (claude-sonnet-4-5-20250929) - Enhanced reasoning and coding capabilities, released September 2025
-- **Claude Opus 4.1** (claude-opus-4-1-20250805) - Most capable model with extended thinking, released August 2025
+- **Claude Haiku 4.5** (claude-haiku-4-5-20251001) - Fast and cost-effective (default), 200K context window
+- **Claude Sonnet 4.6** (claude-sonnet-4-6) - Enhanced reasoning and coding capabilities
+- **Claude Opus 4.6** (claude-opus-4-6) - Most capable model with extended thinking
 
 **Google Gemini Latest Models:**
 
-- **Gemini 3 Pro** (gemini-3-pro) - Latest flagship model with 1M context window, released November 2025
-- **Gemini 3 Pro Preview** (gemini-3-pro-preview-11-2025) - Preview version of Gemini 3
+- **Gemini 3.1 Pro** (gemini-3.1-pro-preview) - Reasoning-first model with 1M context (default)
+- **Gemini 3 Flash** (gemini-3-flash-preview) - Fast with strong coding and reasoning
+- **Gemini 3 Pro** (gemini-3-pro) - Previous flagship (now aliases to 3.1 Pro)
 - **Gemini 2.5 Flash** (gemini-2.5-flash) - Fast and cost-effective
 - **Gemini 2.5 Pro** (gemini-2.5-pro) - Advanced reasoning with thinking mode
 
diff --git a/README.md b/README.md
@@ -45,9 +45,9 @@ Choose the provider that fits your needs and budget:
 
 | Provider          | Best For              | Latest Models                                |
 | ----------------- | --------------------- | -------------------------------------------- |
-| **OpenAI**        | General purpose       | GPT-5.1, GPT-5.1 Codex, GPT-5                |
-| **Anthropic**     | Code understanding    | Claude Haiku 4.5, Sonnet 4.5, Opus 4.1       |
-| **Google Gemini** | Fast responses        | Gemini 3 Pro, Gemini 2.5 Flash               |
+| **OpenAI**        | General purpose       | GPT-5.4, GPT-5.1, GPT-5.1 Codex, GPT-5       |
+| **Anthropic**     | Code understanding    | Claude Haiku 4.5, Sonnet 4.6, Opus 4.6       |
+| **Google Gemini** | Fast responses        | Gemini 3.1 Pro, Gemini 3 Flash                |
 | **Mistral**       | European compliance   | Mistral Small, Mistral Large                 |
 | **DeepSeek**      | Cost-effective        | DeepSeek-V3.1 Chat & Reasoner                |
 | **Qwen**          | Alibaba Cloud         | Qwen Max, Qwen Plus, Qwen Turbo              |
@@ -81,7 +81,7 @@ Access via `Settings` > `Extensions` > `Universal Commit Assistant`:
 
 - **Temperature**: Control creativity (0 = consistent, 2 = creative)
 - **Max Tokens**: Message length limit (100-500)
-- **Max Diff Length**: Maximum characters of git diff to send to AI (1000-10000, default: 3000)
+- **Max Diff Length**: Maximum characters of git diff to send to AI (1000-100000, default: 100000)
 - **Detect First Commit**: Automatically detect and generate appropriate initial commit messages (default: enabled)
 - **Custom Prompt**: Override default instructions
 
@@ -164,18 +164,18 @@ Based on enhanced analysis (~2,500 input tokens + 150 output tokens average):
 
 | Provider             | Model              | Cost per Commit | Cost per 100 Commits |
 | -------------------- | ------------------ | --------------- | -------------------- |
-| **OpenAI**           | GPT-5.1            | ~$0.0008        | ~$0.08               |
+| **OpenAI**           | GPT-5.4            | ~$0.0008        | ~$0.08               |
 | **OpenAI**           | GPT-5.1 Codex      | ~$0.005         | ~$0.50               |
 | **Anthropic**        | Claude Haiku 4.5   | ~$0.003         | ~$0.30               |
-| **Anthropic**        | Claude Sonnet 4.5  | ~$0.010         | ~$1.00               |
+| **Anthropic**        | Claude Sonnet 4.6  | ~$0.010         | ~$1.00               |
 | **DeepSeek**         | deepseek-chat      | ~$0.0009        | ~$0.09               |
-| **Gemini**           | Gemini 3 Pro       | ~$0.0003        | ~$0.03               |
+| **Gemini**           | Gemini 3.1 Pro     | ~$0.0003        | ~$0.03               |
 | **Qwen**             | Qwen Plus          | ~$0.0006        | ~$0.06               |
 | **Ollama/LM Studio** | Local Models       | **FREE**        | **FREE**             |
 
 ### Cost Optimization Tips
 
-1. **Use Cost-Effective Providers**: Gemini 3 Pro, DeepSeek, Qwen, and GPT-5.1 offer excellent quality at low cost
+1. **Use Cost-Effective Providers**: Gemini 3 Pro, DeepSeek, Qwen, and GPT-5.4 offer excellent quality at low cost
 2. **Adjust Max Diff Length**: Reduce `maxDiffLength` to 2000 for simpler commits
 3. **Local Models**: Use Ollama or LM Studio for completely free operation
 4. **Smart Defaults**: The extension uses intelligent truncation to minimize tokens while preserving context
diff --git a/SETTINGS.md b/SETTINGS.md
@@ -109,29 +109,33 @@ Complete configuration guide for customizing Universal Commit Assistant to your
 
 ### OpenAI
 ```json
-"universal-commit-assistant.openai.model": "gpt-5-mini"
+"universal-commit-assistant.openai.model": "gpt-5.4"
 ```
 **Available Models**:
-- `gpt-5-mini`: Fast and cost-effective (recommended)
-- `gpt-5`: Flagship model with state-of-the-art capabilities
+- `gpt-5.4`: Latest model, best intelligence and speed (recommended)
+- `gpt-5.1`: Previous generation, excellent balance
+- `gpt-5.1-codex`: Optimized for coding tasks
+- `gpt-5-mini`: Fast and cost-effective
+- `gpt-5`: Previous flagship model
 
 ### Anthropic
 ```json
 "universal-commit-assistant.anthropic.model": "claude-haiku-4-5-20251001"
 ```
 **Popular Models**:
 - `claude-haiku-4-5-20251001`: Fast and cost-effective (recommended)
-- `claude-sonnet-4-5-20250929`: Enhanced reasoning and coding capabilities
-- `claude-opus-4-1-20250805`: Most capable with extended thinking
+- `claude-sonnet-4-6`: Enhanced reasoning and coding capabilities
+- `claude-opus-4-6`: Most capable with extended thinking
 
 ### Google Gemini
 ```json
-"universal-commit-assistant.gemini.model": "gemini-1.5-flash"
+"universal-commit-assistant.gemini.model": "gemini-3.1-pro-preview"
 ```
 **Popular Models**:
-- `gemini-1.5-flash`: Fast and efficient
-- `gemini-2.0-flash-exp`: Latest experimental model
-- `gemini-1.5-pro`: More capable version
+- `gemini-3.1-pro-preview`: Reasoning-first model with 1M context (recommended)
+- `gemini-3-flash-preview`: Fast with strong coding and reasoning
+- `gemini-3-pro`: Previous flagship (aliases to 3.1 Pro)
+- `gemini-2.5-flash`: Fast and cost-effective
 
 ### Mistral AI
 ```json
@@ -170,10 +174,10 @@ Complete configuration guide for customizing Universal Commit Assistant to your
 
 ### OpenRouter
 ```json
-"universal-commit-assistant.openrouter.model": "openai/gpt-5-mini"
+"universal-commit-assistant.openrouter.model": "openai/gpt-5.4"
 ```
 **Popular Models**:
-- `openai/gpt-5-mini`: OpenAI via OpenRouter
+- `openai/gpt-5.4`: OpenAI via OpenRouter
 - `anthropic/claude-haiku-4-5`: Anthropic via OpenRouter
 - `google/gemini-2.0-flash-exp`: Google via OpenRouter
 - `qwen/qwen-2.5-coder-32b-instruct`: Specialized coding model
@@ -207,15 +211,15 @@ Access via Command Palette (`Ctrl+Shift+P`):
 }
 ```
 
-### Detailed Commits with GPT-5
+### Detailed Commits with GPT-5.4
 ```json
 {
   "universal-commit-assistant.provider": "openai",
   "universal-commit-assistant.messageStyle": "detailed",
   "universal-commit-assistant.temperature": 0.4,
   "universal-commit-assistant.maxTokens": 200,
   "universal-commit-assistant.language": "english",
-  "universal-commit-assistant.openai.model": "gpt-5"
+  "universal-commit-assistant.openai.model": "gpt-5.4"
 }
 ```
 
@@ -237,7 +241,7 @@ Access via Command Palette (`Ctrl+Shift+P`):
   "universal-commit-assistant.messageStyle": "conventional",
   "universal-commit-assistant.language": "chinese",
   "universal-commit-assistant.temperature": 0.3,
-  "universal-commit-assistant.gemini.model": "gemini-1.5-flash"
+  "universal-commit-assistant.gemini.model": "gemini-3.1-pro-preview"
 }
 ```
 
@@ -287,12 +291,12 @@ Access via Command Palette (`Ctrl+Shift+P`):
 
 #### Faster Responses
 - Use local providers (Ollama/LM Studio)
-- Choose faster models (gpt-5-mini, claude-haiku-4-5)
+- Choose faster models (gpt-5.4, claude-haiku-4-5)
 - Reduce maxTokens for shorter messages
 - Set lower temperature for more deterministic output
 
 #### Better Quality
-- Use premium models (gpt-5, claude-sonnet-4-5, claude-opus-4-1)
+- Use premium models (gpt-5.4, claude-sonnet-4-6, claude-opus-4-6)
 - Include unstaged changes for more context
 - Customize system prompt for your coding style
 - Use detailed message style for complex changes
diff --git a/package.json b/package.json
@@ -181,10 +181,10 @@
           },
           "universal-commit-assistant.maxDiffLength": {
             "type": "number",
-            "default": 3000,
+            "default": 100000,
             "minimum": 1000,
-            "maximum": 10000,
-            "description": "Maximum characters of git diff to send to AI (uses smart truncation for larger diffs)",
+            "maximum": 100000,
+            "description": "Maximum characters of git diff to send to AI for deep analysis (uses smart truncation for larger diffs)",
             "order": 8
           },
           "universal-commit-assistant.detectFirstCommit": {
@@ -206,20 +206,22 @@
         "properties": {
           "universal-commit-assistant.openai.model": {
             "type": "string",
-            "default": "gpt-5.1",
+            "default": "gpt-5.4",
             "enum": [
+              "gpt-5.4",
               "gpt-5.1",
               "gpt-5.1-codex",
               "gpt-5.1-codex-mini",
               "gpt-5-mini",
               "gpt-5"
             ],
             "enumDescriptions": [
-              "GPT-5.1 - Latest model balancing intelligence and speed (recommended)",
+              "GPT-5.4 - Latest model, best intelligence and speed (recommended)",
+              "GPT-5.1 - Previous generation, excellent balance",
               "GPT-5.1 Codex - Optimized for coding tasks",
               "GPT-5.1 Codex Mini - Fast coding model",
               "GPT-5 Mini - Fast and cost-effective",
-              "GPT-5 - Full capability flagship model"
+              "GPT-5 - Previous flagship model"
             ],
             "description": "OpenAI model to use"
           },
@@ -228,28 +230,30 @@
             "default": "claude-haiku-4-5-20251001",
             "enum": [
               "claude-haiku-4-5-20251001",
-              "claude-sonnet-4-5-20250929",
-              "claude-opus-4-1-20250805"
+              "claude-sonnet-4-6",
+              "claude-opus-4-6"
             ],
             "enumDescriptions": [
               "Claude Haiku 4.5 - Fast and cost-effective (recommended)",
-              "Claude Sonnet 4.5 - Enhanced reasoning and coding capabilities",
-              "Claude Opus 4.1 - Most capable model with extended thinking"
+              "Claude Sonnet 4.6 - Enhanced reasoning and coding capabilities",
+              "Claude Opus 4.6 - Most capable model with extended thinking"
             ],
             "description": "Anthropic model to use"
           },
           "universal-commit-assistant.gemini.model": {
             "type": "string",
-            "default": "gemini-3-pro",
+            "default": "gemini-3.1-pro-preview",
             "enum": [
+              "gemini-3.1-pro-preview",
+              "gemini-3-flash-preview",
               "gemini-3-pro",
-              "gemini-3-pro-preview-11-2025",
               "gemini-2.5-flash",
               "gemini-2.5-pro"
             ],
             "enumDescriptions": [
-              "Gemini 3 Pro - Latest flagship model with 1M context (recommended)",
-              "Gemini 3 Pro Preview - Preview version of Gemini 3",
+              "Gemini 3.1 Pro - Reasoning-first model with 1M context (recommended)",
+              "Gemini 3 Flash - Fast with strong coding and reasoning",
+              "Gemini 3 Pro - Previous flagship (now aliases to 3.1 Pro)",
               "Gemini 2.5 Flash - Fast and cost-effective",
               "Gemini 2.5 Pro - Advanced reasoning with thinking mode"
             ],
@@ -262,7 +266,7 @@
           },
           "universal-commit-assistant.openrouter.model": {
             "type": "string",
-            "default": "openai/gpt-5-mini",
+            "default": "openai/gpt-5.4",
             "description": "OpenRouter model to use"
           },
           "universal-commit-assistant.deepseek.model": {
diff --git a/src/providers/anthropicProvider.ts b/src/providers/anthropicProvider.ts
@@ -45,8 +45,9 @@ export class AnthropicProvider extends BaseProvider {
           headers: {
             "x-api-key": apiKey,
             "Content-Type": "application/json",
-            "anthropic-version": "2023-06-01",
+            "anthropic-version": "2024-10-22",
           },
+          timeout: 30000,
         }
       );
 
diff --git a/src/providers/deepseekProvider.ts b/src/providers/deepseekProvider.ts
@@ -49,6 +49,7 @@ export class DeepSeekProvider extends BaseProvider {
             Authorization: `Bearer ${apiKey}`,
             "Content-Type": "application/json",
           },
+          timeout: 30000,
         }
       );
 
diff --git a/src/providers/geminiProvider.ts b/src/providers/geminiProvider.ts
@@ -22,6 +22,7 @@ export class GeminiProvider extends BaseProvider {
     const style = options?.style || this.configManager.getMessageStyle();
     const language = this.configManager.getLanguage();
 
+    const maxTokens = options?.maxTokens || (style === "detailed" ? 300 : this.configManager.getMaxTokens());
     const userPrompt = this.buildPrompt(changes, style, options?.customPrompt, language, options?.isFirstCommit);
     const fullPrompt = `${systemPrompt}\n\nUser request: ${userPrompt}\n\nPlease respond with ONLY the commit message, no explanations or additional text.`;
 
@@ -40,13 +41,15 @@ export class GeminiProvider extends BaseProvider {
           ],
           generationConfig: {
             temperature: temperature,
+            maxOutputTokens: maxTokens,
           },
         },
         {
           headers: {
             "x-goog-api-key": apiKey,
             "Content-Type": "application/json",
           },
+          timeout: 30000,
         }
       );
 
diff --git a/src/providers/mistralProvider.ts b/src/providers/mistralProvider.ts
@@ -49,6 +49,7 @@ export class MistralProvider extends BaseProvider {
             Authorization: `Bearer ${apiKey}`,
             "Content-Type": "application/json",
           },
+          timeout: 30000,
         }
       );
 
diff --git a/src/providers/ollamaProvider.ts b/src/providers/ollamaProvider.ts
@@ -15,6 +15,7 @@ export class OllamaProvider extends BaseProvider {
     const systemPrompt = this.configManager.getSystemPrompt();
     const style = options?.style || this.configManager.getMessageStyle();
     const language = this.configManager.getLanguage();
+    const maxTokens = options?.maxTokens || (style === "detailed" ? 300 : this.configManager.getMaxTokens());
 
     const userPrompt = this.buildPrompt(changes, style, options?.customPrompt, language, options?.isFirstCommit);
     const fullPrompt = `${systemPrompt}\n\nUser request: ${userPrompt}\n\nPlease respond with ONLY the commit message, no explanations or additional text.`;
@@ -28,6 +29,7 @@ export class OllamaProvider extends BaseProvider {
           stream: false,
           options: {
             temperature: temperature,
+            num_predict: maxTokens,
           },
         },
         {
diff --git a/src/providers/openaiProvider.ts b/src/providers/openaiProvider.ts
@@ -49,6 +49,7 @@ export class OpenAIProvider extends BaseProvider {
             Authorization: `Bearer ${apiKey}`,
             "Content-Type": "application/json",
           },
+          timeout: 30000,
         }
       );
 
diff --git a/src/providers/openrouterProvider.ts b/src/providers/openrouterProvider.ts
@@ -52,6 +52,7 @@ export class OpenRouterProvider extends BaseProvider {
             "HTTP-Referer": "https://github.com/gianged/universal-commit-assistant",
             "X-Title": "Universal Commit Assistant",
           },
+          timeout: 30000,
         }
       );
 
diff --git a/src/providers/qwenProvider.ts b/src/providers/qwenProvider.ts
@@ -60,6 +60,7 @@ export class QwenProvider extends BaseProvider {
             Authorization: `Bearer ${apiKey}`,
             "Content-Type": "application/json",
           },
+          timeout: 30000,
         }
       );
 
diff --git a/src/services/gitCommitService.ts b/src/services/gitCommitService.ts
diff --git a/src/services/secretManager.ts b/src/services/secretManager.ts
diff --git a/src/types/index.ts b/src/types/index.ts
diff --git a/src/utils/configurationManager.ts b/src/utils/configurationManager.ts
diff --git a/src/utils/retryHandler.ts b/src/utils/retryHandler.ts

Original file line number	Diff line number	Diff line change
`@@ -49,6 +49,7 @@ export class DeepSeekProvider extends BaseProvider {`
`49`	`49`	Authorization: `Bearer ${apiKey}`,
`50`	`50`	`"Content-Type": "application/json",`
`51`	`51`	`},`
	`52`	`+ timeout: 30000,`
`52`	`53`	`}`
`53`	`54`	`);`
`54`	`55`
Original file line number	Diff line number	Diff line change
`@@ -49,6 +49,7 @@ export class MistralProvider extends BaseProvider {`
`49`	`49`	Authorization: `Bearer ${apiKey}`,
`50`	`50`	`"Content-Type": "application/json",`
`51`	`51`	`},`
	`52`	`+ timeout: 30000,`
`52`	`53`	`}`
`53`	`54`	`);`
`54`	`55`
Original file line number	Diff line number	Diff line change
`@@ -49,6 +49,7 @@ export class OpenAIProvider extends BaseProvider {`
`49`	`49`	Authorization: `Bearer ${apiKey}`,
`50`	`50`	`"Content-Type": "application/json",`
`51`	`51`	`},`
	`52`	`+ timeout: 30000,`
`52`	`53`	`}`
`53`	`54`	`);`
`54`	`55`