feat(aihubmix): add Qwen3.7 Max, Gemini 3.5 Flash, Grok 4.3 models and improve Anthropic support

models/aihubmix/manifest.yaml

@@ -1,4 +1,4 @@
-version: 0.0.28
+version: 0.0.29
 type: plugin
 author: langgenius
 name: aihubmix
@@ -8,10 +8,10 @@ label:
   zh_Hans: Aihubmix
   pt_BR: Aihubmix
 description:
-  en_US: 'One Gateway, Infinite Models；one-stop request: OpenAI, Claude, Gemini, DeepSeek, Qwen, and over 500 AI models.'
-  ja_JP: '1つのAPIで無限のAIモデルを利用可能。OpenAI、Claude、Gemini、DeepSeek、Qwenなど、500種類以上のAIモデルにワンストップでアクセス。'
-  zh_Hans: '一个接口，无限模型；一站式请求：OpenAI、Claude、Gemini、DeepSeek、Qwen等 500+AI 模型'
-  pt_BR: 'Um Gateway, Modelos Infinitos; solicitação única: OpenAI, Claude, Gemini, DeepSeek, Qwen e mais de 500 modelos de IA.'
+  en_US: "One Gateway, Infinite Models；one-stop request: OpenAI, Claude, Gemini, DeepSeek, Qwen, and over 500 AI models."
+  ja_JP: "1つのAPIで無限のAIモデルを利用可能。OpenAI、Claude、Gemini、DeepSeek、Qwenなど、500種類以上のAIモデルにワンストップでアクセス。"
+  zh_Hans: "一个接口，无限模型；一站式请求：OpenAI、Claude、Gemini、DeepSeek、Qwen等 500+AI 模型"
+  pt_BR: "Um Gateway, Modelos Infinitos; solicitação única: OpenAI, Claude, Gemini, DeepSeek, Qwen e mais de 500 modelos de IA."
 icon: icon.svg
 resource:
   memory: 268435456

models/aihubmix/models/llm/_position.yaml

@@ -1,3 +1,6 @@
+- qwen3.7-max
+- gemini-3.5-flash
+- grok-4.3
 - gpt-5.5
 - claude-opus-4-7
 - deepseek-v4-flash

models/aihubmix/models/llm/anthropic.py

@@ -201,6 +201,35 @@ def _chat_generate(
                 "max_tokens_to_sample"
             )
 
+        if "max_tokens" not in model_parameters:
+            # Model-aware default: use each model's native max output limit
+            # See: https://platform.claude.com/docs/en/about-claude/models/overview
+            _MAX_TOKENS_128K = {
+                "claude-opus-4-7", "claude-opus-4-7-think",
+                "claude-opus-4-6", "claude-opus-4-6-think",
+            }
+            _MAX_TOKENS_64K = {
+                "claude-sonnet-4-6", "claude-sonnet-4-6-think",
+                "claude-sonnet-4-5",
+                "claude-sonnet-4-20250514",
+                "claude-haiku-4-5",
+                "claude-opus-4-5", "claude-opus-4-5-think",
+                "claude-opus-4-1-20250805",
+                "claude-opus-4-20250514",
+            }
+            _MAX_TOKENS_8K = {
+                "claude-3-5-sonnet-20241022",
+                "claude-3-5-haiku-20241022",
+            }
+            if model in _MAX_TOKENS_128K:
+                model_parameters["max_tokens"] = 128000
+            elif model in _MAX_TOKENS_64K:
+                model_parameters["max_tokens"] = 64000
+            elif model in _MAX_TOKENS_8K:
+                model_parameters["max_tokens"] = 8192
+            else:
+                model_parameters["max_tokens"] = 4096
+
         thinking = model_parameters.pop("thinking", False)
         thinking_budget = model_parameters.pop("thinking_budget", 1024)
         context_1m = model_parameters.pop("context_1m", False)

models/aihubmix/models/llm/claude-haiku-4-5.yaml

@@ -37,9 +37,9 @@ parameter_rules:
     use_template: top_p
   - name: max_tokens
     use_template: max_tokens
-    default: 8192
+    default: 64000
     min: 1
-    max: 131072
+    max: 64000
   - name: response_format
     label:
       zh_Hans: 回复格式

models/aihubmix/models/llm/claude-opus-4-5.yaml

@@ -18,9 +18,9 @@ parameter_rules:
     use_template: top_p
   - name: max_tokens
     use_template: max_tokens
-    default: 8192
+    default: 64000
     min: 1
-    max: 32000
+    max: 64000
   - name: response_format
     label:
       zh_Hans: 回复格式

models/aihubmix/models/llm/claude-opus-4-7.yaml

@@ -18,9 +18,9 @@ parameter_rules:
     use_template: top_p
   - name: max_tokens
     use_template: max_tokens
-    default: 8192
+    default: 128000
     min: 1
-    max: 32000
+    max: 128000
   - name: response_format
     label:
       zh_Hans: 回复格式

models/aihubmix/models/llm/gemini-3.5-flash.yaml

@@ -0,0 +1,134 @@
+model: gemini-3.5-flash
+label:
+  en_US: Gemini 3.5 Flash
+model_type: llm
+features:
+  - tool-call
+  - multi-tool-call
+  - agent-thought
+  - vision
+  - stream-tool-call
+  - document
+  - video
+  - audio
+  - structured-output
+model_properties:
+  mode: chat
+  context_size: 1048576
+parameter_rules:
+  - name: include_thoughts
+    type: boolean
+    required: true
+    default: false
+    label:
+      en_US: Include thoughts
+      zh_Hans: 返回思考过程
+    help:
+      en_US: Indicates whether to include thoughts in the response. If true, thoughts are returned only if the model supports thought and thoughts are available. Default is false.
+      zh_Hans: 是否返回思考过程。若为 true 则只有模型支持思考且启动思考模式时才返回思考过程。默认为 false。
+  - name: media_resolution
+    type: string
+    required: true
+    default: "Default"
+    options:
+      - Default
+      - Low
+      - Medium
+      - High
+    label:
+      en_US: Media resolution
+      zh_Hans: 媒体分辨率
+    help:
+      en_US: |
+        Gemini 3 introduces fine-grained control over multimodal visual processing via the `media_resolution` parameter. Higher resolution improves the model's ability to read small text or recognize fine details, but increases token usage and latency.
+
+        Recommended Settings:
+        - Images: `media_resolution_high` (1120 tokens) - Recommended for most image analysis tasks to ensure best quality.
+        - PDF: `media_resolution_medium` (560 tokens) - Great for document understanding; quality usually saturates at medium.
+        - Video (Regular): `media_resolution_low` or `medium` (70 tokens/frame) - Sufficient for most action recognition and description tasks.
+        - Video (Text-heavy): `media_resolution_high` (280 tokens/frame) - Needed only when use cases involve reading dense text (OCR) or tiny details in video frames.
+
+        Note: If not specified, the model uses the best default value based on the media type.
+      zh_Hans: |
+        Gemini 3 通过 `media_resolution` 参数引入了对多模态视觉处理的精细控制。分辨率越高，模型读取细小文本或识别细微细节的能力就越强，但会增加令牌用量和延迟时间。
+
+        推荐设置：
+        - 图片：`media_resolution_high` (1120 tokens) - 建议用于大多数图片分析任务，以确保获得最佳质量。
+        - PDF：`media_resolution_medium` (560 tokens) - 非常适合文档理解；质量通常在 medium 时达到饱和。
+        - 视频（常规）：`media_resolution_low` 或 `medium` (70 tokens/帧) - 对于大多数动作识别和描述任务来说已经足够。
+        - 视频（文字较多）：`media_resolution_high` (280 tokens/帧) - 仅当用例涉及读取密集文本 (OCR) 或视频帧中的微小细节时才需要。
+
+        注意：如果不指定，模型会根据媒体类型使用最佳默认值。
+  - name: thinking_level
+    type: string
+    required: true
+    default: "Medium"
+    options:
+      - Minimal
+      - Low
+      - Medium
+      - High
+    label:
+      en_US: Thinking level
+      zh_Hans: 思考层级
+    help:
+      en_US: Indicates the thinking level. Default is Medium.
+      zh_Hans: 思考层级。默认为 Medium。
+  - name: grounding
+    type: boolean
+    required: true
+    default: false
+    label:
+      en_US: Grounding
+      zh_Hans: 事实核查
+      ja_JP: 事実チェック
+    help:
+      en_US: Grounding with Google Search
+      zh_Hans: Google 事实核查
+      ja_JP: Google 検索に基づいた応答をします。
+  - name: url_context
+    type: boolean
+    required: true
+    default: false
+    label:
+      en_US: URL context
+      zh_Hans: 链接详情
+      ja_JP: 網羅閲覧
+    help:
+      en_US: Browse the url context
+      zh_Hans: 获取并分析该链接的详细信息，为回答提供参考依据。
+      ja_JP: リンク先の詳細情報を取得し、文脈を理解します。
+  - name: code_execution
+    type: boolean
+    required: true
+    default: false
+    label:
+      en_US: Code execution
+      zh_Hans: 代码执行
+      ja_JP: コード実行
+    help:
+      en_US: Lets Gemini use code to solve complex tasks
+      zh_Hans: 让 Gemini 使用代码来解决复杂任务。
+      ja_JP: Gemini にコードを使って複雑なタスクを解決させましょう。
+  - name: use_inline_file
+    type: boolean
+    required: true
+    default: false
+    label:
+      en_US: Use inline file mode
+      zh_Hans: 使用内嵌文件模式
+    help:
+      en_US: |
+        When enabled, files will be embedded directly in the request (inline mode) for better performance.
+        When disabled, files will be uploaded using the Files API. Default is disabled (Files API).
+      zh_Hans: |
+        启用后，文件将直接内嵌在请求中（内联模式）以提高性能。
+        禁用后，文件将使用 Files API 上传。默认为禁用（使用 Files API）。
+  - name: json_schema
+    use_template: json_schema
+# https://ai.google.dev/gemini-api/docs/pricing#gemini-3-flash-preview
+pricing:
+  input: "1.5"
+  output: "9"
+  unit: "0.000001"
+  currency: USD

models/aihubmix/models/llm/grok-4.3.yaml

@@ -0,0 +1,46 @@
+model: grok-4.3
+label:
+  en_US: Grok 4.3
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+  - stream-tool-call
+  - multi-tool-call
+  - vision
+model_properties:
+  mode: chat
+  context_size: 1000000
+parameter_rules:
+  - name: reasoning_effort
+    label:
+      zh_Hans: 推理努力程度
+      en_US: Reasoning Effort
+    type: string
+    help:
+      zh_Hans: 控制模型的推理强度。none 为关闭思考，low/medium/high 为推荐值。
+      en_US: Controls the model's reasoning intensity. none disables thinking, low/medium/high are recommended.
+    required: false
+    default: none
+    options:
+      - none
+      - low
+      - medium
+      - high
+  - name: response_format
+    label:
+      zh_Hans: 回复格式
+      en_US: Response Format
+    type: string
+    required: false
+    options:
+      - text
+      - json_object
+      - json_schema
+  - name: json_schema
+    use_template: json_schema
+pricing:
+  input: "1.25"
+  output: "2.5"
+  unit: "0.000001"
+  currency: USD

models/aihubmix/models/llm/qwen3.7-max.yaml

@@ -0,0 +1,37 @@
+model: qwen3.7-max
+label:
+  en_US: Qwen3.7 Max
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+  - stream-tool-call
+  - multi-tool-call
+model_properties:
+  mode: chat
+  context_size: 1014784
+parameter_rules:
+  - name: enable_thinking
+    required: false
+    type: boolean
+    default: false
+    label:
+      zh_Hans: 深度思考
+      en_US: Deep Thinking
+    help:
+      zh_Hans: 是否开启深度思考模式，开启后模型会先进行推理再给出回答。
+      en_US: Whether to enable deep thinking mode. When enabled, the model will reason before answering.
+  - name: response_format
+    label:
+      zh_Hans: 回复格式
+      en_US: Response Format
+    type: string
+    required: false
+    options:
+      - text
+      - json_object
+pricing:
+  input: "1.69"
+  output: "5.07"
+  unit: "0.000001"
+  currency: USD