docs: add Qwen3.7-Max and Qwen3.8-Max

docs/llmservice/models/qwen3.7-max.md

@@ -0,0 +1,46 @@
+# Qwen3.7-Max
+
+Qwen3.7-Max is a Qwen model available through B.AI for programming, office productivity, long-context analysis, and tool-assisted workflows. On B.AI, use the model ID `qwen3.7-max`.
+
+## Key Features
+
+* **Hybrid Thinking:** Supports thinking and non-thinking response modes where available, allowing different depth and latency tradeoffs by task.
+* **Long Context:** Supports up to 1,000,000 tokens of context, subject to B.AI platform configuration and request limits.
+* **Agent and Tool Workflows:** Suitable for function calling, structured output, and tool-assisted tasks where these capabilities are enabled.
+* **Context Caching:** Supports cache-aware usage. Cache behavior and availability are subject to the current B.AI model configuration.
+
+## Best Use Cases
+
+* Repository-scale implementation, debugging, code review, and multi-step engineering tasks.
+* Large-document analysis, research synthesis, and enterprise knowledge workflows.
+* Document drafting, summarization, data analysis, and structured productivity tasks.
+* Tool-driven automation that uses function calling or structured output.
+
+## Capabilities and Limitations
+
+| Capability | Description |
+| :--- | :--- |
+| **Reasoning** | Supports thinking and non-thinking modes where available. |
+| **Coding** | Suitable for code generation, debugging, review, refactoring, and multi-step engineering workflows. |
+| **Creative Writing** | Supports general-purpose drafting and open-ended text generation. |
+| **Multimodal** | The `qwen3.7-max` model ID accepts text input and returns text output. |
+| **Context Window** | Up to 1,000,000 tokens, subject to platform configuration and request limits. |
+| **Max Output** | Up to 65,536 tokens, subject to platform configuration. |
+| **Tool Use** | Function calling and structured output are available where enabled. |
+| **Multilingual** | Suitable for multilingual text-generation workflows. |
+
+### Known Limitations
+
+* Model availability, input modalities, tool support, and request limits depend on the current B.AI model catalog and platform configuration.
+* The model ID may be updated by the provider or platform. Verify the current model catalog before relying on fixed behavior in production workflows.
+* Model output can be inaccurate. Validate results before using them in critical decisions or production systems.
+
+## Pricing
+
+| Model | Input (Credits/Token) | Cache Write (Credits/Token) | Cache Read (Credits/Token) | Output (Credits/Token) | Web Search (Credits/Use) | Billing Notes |
+| :--- | --------------------: | --------------------------: | -------------------------: | ---------------------: | -----------------------: | :--- |
+| **Qwen3.7-Max** | `1.65` | `1.65` | `0.33` | `4.951` | `-` | - |
+
+:::info Pricing note
+Prices shown in the documentation are B.AI standard reference prices for base billing purposes. B.AI may provide lower actual usage costs through top-up bonuses and account benefits. Specific prices, bonus Credits, and account benefits are subject to the platform display and final billing records.
+:::

docs/llmservice/models/qwen3.8-max.md

@@ -0,0 +1,53 @@
+# Qwen3.8-Max
+
+## Overview
+
+Qwen3.8-Max is Alibaba's flagship Qwen model for coding, professional knowledge work, multimodal understanding, and long-horizon agent tasks. It was officially released on August 3, 2026, with the API model ID `qwen3.8-max`. The model uses a Mixture-of-Experts (MoE) architecture with 2.4 trillion total parameters and 95 billion activated parameters, accepts text, image, and video input, and produces text output.
+
+## Key Features
+
+* **2.4T-Parameter MoE Architecture:** Built on the Qwen3.5 architectural foundation with 2.4 trillion total parameters and 95 billion activated parameters per token.
+* **Native Multimodal Intelligence:** Processes text, images, and video in the same model. Qwen describes vision as part of an agent's planning, execution, and self-correction loop rather than a separate preprocessing step.
+* **1M-Token Context:** Provides a 1-million-token context window, with up to 991.80K input tokens in non-thinking mode, 983.61K input tokens in thinking mode, and 131.07K output tokens.
+* **Configurable Reasoning:** Thinking is enabled by default. The API supports `low`, `medium`, and `xhigh` reasoning effort, with `xhigh` as the default, and preserves reasoning across turns by default.
+* **Long-Horizon Agent Work:** The launch report documents autonomous coding, research, chip-design optimization, and year-long business-simulation tasks. In one coding case, the model operated for about 16 days and produced 265 commits, 127 pull requests, and 151 issues without human intervention.
+* **Agent and API Integration:** Supports Function Calling, structured output, prefix completion, batch inference, context caching, and built-in tools. It can be accessed through OpenAI-compatible Chat Completions and Responses APIs, Anthropic-compatible APIs, or DashScope.
+
+## Best Use Cases
+
+* **Autonomous Software Engineering:** Multi-day implementation, repository maintenance, testing, debugging, and iterative delivery using coding agents.
+* **Research and Experimentation:** Reproducing papers, writing experimental pipelines, running iterative evaluations, and refining hypotheses with tools.
+* **Professional Knowledge Work:** Legal review, financial analysis, document production, data analysis, design prototyping, and other multi-step office workflows.
+* **Long-Horizon Planning and Optimization:** Tasks that require hundreds of feedback cycles, persistent state, adaptive planning, and repeated verification.
+* **Multimodal Agents:** Workflows that combine documents, screenshots, images, video, code execution, search, and external tools.
+
+## Capabilities and Limitations
+
+| Capability | Description |
+| :--- | :--- |
+| **Reasoning** | Thinking is enabled by default. `reasoning_effort` supports `low`, `medium`, and `xhigh`; `xhigh` is the default. The maximum mapped thinking budget is 262,144 tokens, while the default budget is 131,072. |
+| **Coding** | Designed for autonomous, long-horizon coding. Qwen reports 86.6 on Terminal-Bench 2.1, 67.7 on SWE-bench Pro, 73.5 on FrontierSWE, and 93.0 on PaperBench. |
+| **Multimodal** | Accepts text, image, and video input and produces text output. |
+| **Context Window** | 1M tokens. |
+| **Maximum Input** | 991.80K tokens in non-thinking mode and 983.61K tokens in thinking mode. |
+| **Max Output** | 131.07K tokens in both modes. |
+| **Structured Output** | Supported. |
+| **Architecture** | Mixture-of-Experts architecture based on Qwen3.5. |
+| **Total Parameters** | 2.4 trillion. |
+| **Activated Parameters** | 95 billion. |
+| **Knowledge Cutoff** | Unpublished. |
+
+### Known Limitations
+
+* Preserved thinking is enabled by default. Multi-turn clients must return the complete, unmodified `reasoning_content` history; preserved reasoning also counts toward input tokens and billing.
+* Alibaba has not published a model-specific knowledge cutoff, measured generation speed, or complete language-coverage list.
+
+## Pricing
+
+| Model | Input (Credits/Token) | Cache Write (Credits/Token) | Cache Read (Credits/Token) | Output (Credits/Token) | Web Search (Credits/Use) | Billing Notes |
+| :--- | --------------------: | --------------------------: | -------------------------: | ---------------------: | -----------------------: | :--- |
+| **Qwen3.8-Max** | `2.00` | `2.00` | `0.25` | `6.00` | `-` | Cache Read uses the official implicit-cache price. |
+
+:::info Pricing note
+Prices shown in the documentation are B.AI standard reference prices for base billing purposes. B.AI may provide lower actual usage costs through top-up bonuses and account benefits. Specific prices, bonus Credits, and account benefits are subject to the platform display and final billing records.
+:::

docs/llmservice/pricing-and-usage.md

@@ -19,6 +19,8 @@ The platform uses a unified Credits system to measure and settle usage across al
 | Kimi K3           |                  3.00 |                        3.00 |                       0.30 |                  15.00 |                        - |
 | Kimi K2.6         |                  0.95 |                        0.95 |                       0.16 |                   4.00 |                        - |
 | Kimi K2.5         |                  0.59 |                        0.59 |                      0.177 |                   3.00 |                        - |
+| Qwen3.8-Max       |                  2.00 |                        2.00 |                       0.25 |                   6.00 |                        - |
+| Qwen3.7-Max       |                  1.65 |                        1.65 |                       0.33 |                  4.951 |                        - |
 | Qwen3.6-27B       |                  0.19 |                        0.19 |                      0.019 |                   2.99 |                        - |
 | GLM-5.2           |                  1.40 |                        1.40 |                       0.28 |                   4.40 |                        - |
 | GLM-5.1           |                  1.40 |                        1.40 |                       0.26 |                   4.40 |                        - |

i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/qwen3.7-max.md

@@ -0,0 +1,46 @@
+# Qwen3.7-Max
+
+Qwen3.7-Max 是 B.AI 提供的 Qwen 模型，适用于编程、办公效率、长上下文分析和工具辅助工作流。在 B.AI 中使用模型 ID：`qwen3.7-max`。
+
+## 核心特性
+
+* **Hybrid Thinking：** 在可用场景下支持 thinking 与 non-thinking 两种响应模式，可根据任务调整推理深度与响应延迟。
+* **长上下文：** 最多支持 1,000,000 tokens 上下文，具体以 B.AI 平台配置和请求限制为准。
+* **智能体与工具工作流：** 在相关能力启用时，适用于函数调用、结构化输出和工具辅助任务。
+* **上下文缓存：** 支持缓存相关用量；缓存行为和可用性以 B.AI 当前模型配置为准。
+
+## 适用场景
+
+* 仓库级实现、调试、代码审查和多步骤工程任务。
+* 大型文档分析、研究综合和企业知识工作流。
+* 文档起草、内容总结、数据分析和结构化办公任务。
+* 使用函数调用或结构化输出的工具驱动自动化工作流。
+
+## 能力与限制
+
+| 能力维度 | 说明 |
+| :--- | :--- |
+| **推理能力** | 在可用场景下支持 thinking 与 non-thinking 两种模式。 |
+| **编程能力** | 适合代码生成、调试、审查、重构和多步骤工程工作流。 |
+| **创意写作** | 支持通用内容起草和开放式文本生成。 |
+| **多模态能力** | `qwen3.7-max` 模型 ID 支持文本输入和文本输出。 |
+| **上下文窗口** | 最多 1,000,000 tokens，具体以平台配置和请求限制为准。 |
+| **最大输出** | 最多 65,536 tokens，具体以平台配置为准。 |
+| **工具调用** | 在启用时支持函数调用和结构化输出。 |
+| **多语言** | 适用于多语言文本生成工作流。 |
+
+### 已知限制
+
+* 模型可用性、输入模态、工具支持和请求限制取决于 B.AI 当前模型目录及平台配置。
+* 模型 ID 可能会随提供方或平台调整；在生产工作流中依赖固定行为前，请确认当前模型目录。
+* 模型输出可能不准确；关键决策或生产系统使用前应进行校验。
+
+## 价格
+
+| 模型名称 | 输入（Credits/Token） | 缓存写入（Credits/Token） | 缓存读取（Credits/Token） | 输出（Credits/Token） | 网页搜索（Credits/次） | 计费说明 |
+| :--- | --------------------: | -------------------------: | -------------------------: | --------------------: | ---------------------: | :--- |
+| **Qwen3.7-Max** | `1.65` | `1.65` | `0.33` | `4.951` | `-` | - |
+
+:::info 价格说明
+文档价格为 B.AI 平台模型标准参考价，仅供基础计费说明使用。B.AI 可能会通过充值赠送及账户权益等方式，为用户提供更低的实际使用成本。具体价格、赠送积分及账户权益请以平台页面展示及最终账单为准。
+:::

i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/qwen3.8-max.md

@@ -0,0 +1,53 @@
+# Qwen3.8-Max
+
+## 概述
+
+Qwen3.8-Max 是阿里巴巴面向编程、专业知识工作、多模态理解和长周期智能体任务推出的旗舰 Qwen 模型。该模型于 2026 年 8 月 3 日正式发布，API 模型 ID 为 `qwen3.8-max`。模型采用 Mixture-of-Experts（MoE）架构，总参数量为 2.4 万亿，每个 token 激活 950 亿参数；支持文本、图像和视频输入，输出为文本。
+
+## 核心特性
+
+* **2.4T 参数 MoE 架构：** 基于 Qwen3.5 架构基础构建，总参数量为 2.4 万亿，每个 token 激活 950 亿参数。
+* **原生多模态智能：** 在同一模型中处理文本、图像和视频。Qwen 将视觉能力视为智能体规划、执行和自我纠错闭环的一部分，而非独立的预处理步骤。
+* **1M Token 上下文：** 提供 100 万 token 上下文窗口；non-thinking 模式最多支持 991.80K 输入 token，thinking 模式最多支持 983.61K 输入 token，两种模式的最大输出均为 131.07K token。
+* **可配置推理：** 默认开启 thinking。API 支持 `low`、`medium` 和 `xhigh` 三档 `reasoning_effort`，默认使用 `xhigh`，并默认跨轮保留推理内容。
+* **长周期智能体工作：** 发布资料展示了自主编程、研究、芯片设计优化和年度业务模拟等任务。在一个编程案例中，模型连续运行约 16 天，无人工干预地产生了 265 个 commit、127 个 pull request 和 151 个 issue。
+* **智能体与 API 集成：** 支持 Function Calling、结构化输出、前缀补全、批量推理、上下文缓存和内置工具；可通过 OpenAI 兼容的 Chat Completions 和 Responses API、Anthropic 兼容 API 或 DashScope 调用。
+
+## 适用场景
+
+* **自主软件工程：** 使用编程智能体进行多日实现、仓库维护、测试、调试和迭代交付。
+* **研究与实验：** 复现论文、编写实验管线、运行迭代评估，并借助工具持续完善假设。
+* **专业知识工作：** 法律审查、财务分析、文档生产、数据分析、设计原型及其他多步骤办公工作流。
+* **长周期规划与优化：** 需要数百次反馈循环、持久状态、自适应规划和重复验证的任务。
+* **多模态智能体：** 结合文档、截图、图像、视频、代码执行、搜索和外部工具的工作流。
+
+## 能力与限制
+
+| 能力维度 | 说明 |
+| :--- | :--- |
+| **推理能力** | 默认开启 thinking。`reasoning_effort` 支持 `low`、`medium` 和 `xhigh`，默认使用 `xhigh`。最大映射推理预算为 262,144 token，默认预算为 131,072 token。 |
+| **编程能力** | 面向自主、长周期编程设计。Qwen 报告的 Terminal-Bench 2.1 得分为 86.6、SWE-bench Pro 为 67.7、FrontierSWE 为 73.5、PaperBench 为 93.0。 |
+| **多模态能力** | 支持文本、图像和视频输入，输出为文本。 |
+| **上下文窗口** | 100 万 token。 |
+| **最大输入** | non-thinking 模式为 991.80K token，thinking 模式为 983.61K token。 |
+| **最大输出** | 两种模式均为 131.07K token。 |
+| **结构化输出** | 支持。 |
+| **架构** | 基于 Qwen3.5 的 Mixture-of-Experts 架构。 |
+| **总参数量** | 2.4 万亿。 |
+| **激活参数量** | 950 亿。 |
+| **知识截止日期** | 未公开。 |
+
+### 已知限制
+
+* 默认开启跨轮推理保留。多轮客户端必须返回完整且未经修改的 `reasoning_content` 历史；保留的推理内容同样会计入输入 token 和计费。
+* 阿里巴巴尚未公布模型专属的知识截止日期、实测生成速度或完整的语言覆盖列表。
+
+## 价格
+
+| 模型名称 | 输入（Credits/Token） | 缓存写入（Credits/Token） | 缓存读取（Credits/Token） | 输出（Credits/Token） | 网页搜索（Credits/次） | 计费说明 |
+| :--- | --------------------: | -------------------------: | -------------------------: | --------------------: | ---------------------: | :--- |
+| **Qwen3.8-Max** | `2.00` | `2.00` | `0.25` | `6.00` | `-` | 缓存读取采用官方隐式缓存价格。 |
+
+:::info 价格说明
+文档价格为 B.AI 平台模型标准参考价，仅供基础计费说明使用。B.AI 可能会通过充值赠送及账户权益等方式，为用户提供更低的实际使用成本。具体价格、赠送积分及账户权益请以平台页面展示及最终账单为准。
+:::

i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/pricing-and-usage.md

@@ -19,6 +19,8 @@
 | Kimi K3 | 3.00 | 3.00 | 0.30 | 15.00 | - |
 | Kimi K2.6 | 0.95 | 0.95 | 0.16 | 4.00 | - |
 | Kimi K2.5 | 0.59 | 0.59 | 0.177 | 3.00 | - |
+| Qwen3.8-Max | 2.00 | 2.00 | 0.25 | 6.00 | - |
+| Qwen3.7-Max | 1.65 | 1.65 | 0.33 | 4.951 | - |
 | Qwen3.6-27B | 0.19 | 0.19 | 0.019 | 2.99 | - |
 | GLM-5.2 | 1.40 | 1.40 | 0.28 | 4.40 | - |
 | GLM-5.1 | 1.40 | 1.40 | 0.26 | 4.40 | - |

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@x402-tron/docs",
-  "version": "1.3.14",
+  "version": "1.3.15",
   "description": "x402-tron documentation",
   "license": "MIT",
   "resolutions": {

sidebars.js

@@ -236,6 +236,8 @@ const sidebars = {
               label: 'Qwen',
               collapsed: true,
               items: [
+                { type: 'doc', id: 'llmservice/models/qwen3.8-max', label: 'Qwen3.8-Max' },
+                { type: 'doc', id: 'llmservice/models/qwen3.7-max', label: 'Qwen3.7-Max' },
                 { type: 'doc', id: 'llmservice/models/qwen3.6-27b', label: 'Qwen3.6-27B' },
               ],
             },