From 2d89b9be8f9af42307a63af36d4a4bc9f7df7f67 Mon Sep 17 00:00:00 2001 From: ai-bankofai Date: Mon, 30 Mar 2026 14:07:41 +0800 Subject: [PATCH 1/5] Create deepseek-v3.2.md --- docs/llmservice/models/deepseek-v3.2.md | 46 +++++++++++++++++++++++++ 1 file changed, 46 insertions(+) create mode 100644 docs/llmservice/models/deepseek-v3.2.md diff --git a/docs/llmservice/models/deepseek-v3.2.md b/docs/llmservice/models/deepseek-v3.2.md new file mode 100644 index 00000000..9fdfb098 --- /dev/null +++ b/docs/llmservice/models/deepseek-v3.2.md @@ -0,0 +1,46 @@ +# DeepSeek-V3.2 Model Overview + +**DeepSeek-V3.2** is a **reasoning-first** large language model launched by DeepSeek AI, specifically built for **Agentic scenarios**. By implementing the innovative **DeepSeek Sparse Attention (DSA)** mechanism, it achieves exceptional computational efficiency while demonstrating superior performance in complex tasks and autonomous agent capabilities. + +--- + +## 🚀 Key Features + +* **Leading Reasoning Capabilities**: DeepSeek-V3.2 excels across multiple benchmarks, particularly in **Agentic Capabilities**. Its **Thinking Mode** provides high-tier performance in coding and logic tasks, comparable to top-tier models in the industry. +* **DeepSeek Sparse Attention (DSA)**: Optimizes long-context reasoning speed and efficiency, allowing for faster response times even with massive inputs. +* **Optimized for Agents**: + * **Environment Simulation**: Trained on 1,800+ diverse environments. + * **Complex Instruction Following**: Refined with 85k+ sophisticated instructions. + * **Native Tool Integration**: Supports seamless tool calling in both Thinking and non-Thinking modes. +* **Massive Context Support**: Features a **164K token context window**, making it ideal for processing ultra-long documents and multi-turn interactions. + +--- + +## 💡 Best Use Cases + +1. **AI Agent Development**: Building autonomous agents capable of complex planning, multi-step execution, and external tool utilization. +2. **Advanced Coding Assistance**: Providing deep support for software architecture, code review, and automated programming. +3. **Long-Form Text Analysis**: Efficiently processing legal documents, research papers, and technical manuals. +4. **Complex Problem Solving**: Delivering high-quality solutions for tasks requiring deep logical reasoning and chain-of-thought analysis. + +--- + +## 📊 Capabilities and Limitations + +| Capability | Detailed Description | +| :--- | :--- | +| **Reasoning Ability** | **Extremely Strong**. Excels in logical chains and Agent task planning. | +| **Creative Ability** | **Strong**. Proficient in code generation and complex task-flow design. | +| **Multimodal Ability** | Primarily text-based; achieves multimodal tasks via Agent frameworks. | +| **Context Window** | **164K Tokens** (Supports massive document ingestion) | +| **Max Output** | **164K Tokens** (Industry-leading generation limit) | + +--- + +## 💰 Credits and Pricing + +> **Note:** Consumption is calculated per 1,000 Tokens. + +| Model | Input (Credits/1K Tokens) | Output (Credits/1K Tokens) | +| :--- | :---: | :---: | +| **DeepSeek-V3.2** | `0.27` | `0.42` | From f24ae668f1dd5d03a9b31ebfffb9786464462f8d Mon Sep 17 00:00:00 2001 From: ai-bankofai Date: Mon, 30 Mar 2026 14:12:34 +0800 Subject: [PATCH 2/5] Create deepseek-v3.2.md --- .../llmservice/models/deepseek-v3.2.md | 46 +++++++++++++++++++ 1 file changed, 46 insertions(+) create mode 100644 i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/deepseek-v3.2.md diff --git a/i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/deepseek-v3.2.md b/i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/deepseek-v3.2.md new file mode 100644 index 00000000..34ef1c26 --- /dev/null +++ b/i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/deepseek-v3.2.md @@ -0,0 +1,46 @@ +# DeepSeek-V3.2 Model Overview + +**DeepSeek-V3.2** 是由 DeepSeek AI 推出的**推理导向型 (Reasoning-first)** 大语言模型，专为 **智能体 (Agent)** 场景量身打造。该模型通过创新的 **DSA (DeepSeek Sparse Attention)** 稀疏注意力机制极大优化了计算效率，在复杂任务处理与自主代理能力上表现卓越。 + +--- + +## 🚀 核心特性 (Key Features) + +* **顶尖推理性能**：在多项基准测试中表现出色，尤其在 **Agentic Capabilities** 方面。其“思维模式 (Thinking mode)”在编程任务中可比肩顶级模型。 +* **DSA 稀疏注意力机制**：引入自主研发的 **DeepSeek Sparse Attention (DSA)**，显著提升长上下文推理的效率与响应速度。 +* **深度优化 Agent 场景**： + * **海量环境模拟**：覆盖 1800+ 虚拟环境。 + * **复杂指令遵循**：基于 85k+ 复杂指令进行强化训练。 + * **原生工具集成**：支持在“思考”与“非思考”模式下直接调用外部工具。 +* **超长上下文支持**：具备 **164K tokens** 的上下文窗口，可轻松驾驭海量信息处理。 + +--- + +## 💡 最佳应用场景 (Best Use Cases) + +1. **AI Agent 开发**：构建具备复杂规划、多步执行及自主工具调用能力的智能体。 +2. **高级代码辅助**：深度参与软件架构设计、代码审查及自动化编程任务。 +3. **长文本分析**：高效处理法律文档、科研论文及技术手册等超长文本。 +4. **深度逻辑推理**：解决需要严密逻辑链条的高难度科研或业务问题。 + +--- + +## 📊 技术指标与能力边界 (Capabilities & Limitations) + +| 能力维度 | 详细描述 | +| :--- | :--- | +| **推理能力** | **极强**。擅长复杂逻辑链、Agent 任务规划。 | +| **创作能力** | **强**。精通代码生成、文案创作及复杂的任务流设计。 | +| **多模态能力** | 以文本为主，可通过 Agent 框架协同多模态工具实现。 | +| **上下文窗口** | **164K Tokens** (支持超长文本输入) | +| **最大输出** | **164K Tokens** (单次生成上限极高) | + +--- + +## 💰 计费与抵扣说明 (Credits & Pricing) + +> **注意：** 消耗以每 1,000 Tokens 为基准进行折算。 + +| 模型名称 | 输入 (Credits/1K Tokens) | 输出 (Credits/1K Tokens) | +| :--- | :---: | :---: | +| **DeepSeek-V3.2** | `0.27` | `0.42` | From c2c33097fcdebfa6ea24eaae7c269c90a9eb08c0 Mon Sep 17 00:00:00 2001 From: ai-bankofai Date: Mon, 30 Mar 2026 14:22:45 +0800 Subject: [PATCH 3/5] Update deepseek-v3.2.md --- .../current/llmservice/models/deepseek-v3.2.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/deepseek-v3.2.md b/i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/deepseek-v3.2.md index 34ef1c26..ac5f6365 100644 --- a/i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/deepseek-v3.2.md +++ b/i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/deepseek-v3.2.md @@ -1,4 +1,4 @@ -# DeepSeek-V3.2 Model Overview +# DeepSeek-V3.2 概览 **DeepSeek-V3.2** 是由 DeepSeek AI 推出的**推理导向型 (Reasoning-first)** 大语言模型，专为 **智能体 (Agent)** 场景量身打造。该模型通过创新的 **DSA (DeepSeek Sparse Attention)** 稀疏注意力机制极大优化了计算效率，在复杂任务处理与自主代理能力上表现卓越。 From c27d6a75495c54ce01553d8807cc011194d42cdc Mon Sep 17 00:00:00 2001 From: jerryji-prog Date: Mon, 30 Mar 2026 16:20:36 +0800 Subject: [PATCH 4/5] add deepseek-v3.2 --- .github/workflows/docker.yml | 13 ++--- .github/workflows/release-publish.yml | 54 +++++++++++++++++++ docs/llmservice/models/deepseek-v3.2.md | 4 +- .../llmservice/models/deepseek-v3.2.md | 4 +- .../current/sidebars.js | 1 + package.json | 2 +- sidebars.js | 1 + 7 files changed, 66 insertions(+), 13 deletions(-) diff --git a/.github/workflows/docker.yml b/.github/workflows/docker.yml index 0882e992..09ac7358 100644 --- a/.github/workflows/docker.yml +++ b/.github/workflows/docker.yml @@ -5,6 +5,7 @@ on: branches: - main - master + - update-mcp-server - ai-bankofai-patch-1 tags: - 'test' @@ -12,8 +13,9 @@ on: branches: - main - master + - update-mcp-server - ai-bankofai-patch-1 - workflow_dispatch: + workflow_dispatch: # Intentionally unrestricted — allows manual builds from any branch for flexibility env: IMAGE_NAME: bankofai/docs @@ -31,15 +33,6 @@ jobs: - name: Set up Docker Buildx uses: docker/setup-buildx-action@v3 - - name: Debug Docker Hub secrets - if: github.event_name != 'pull_request' - env: - DOCKERHUB_USERNAME: ${{ secrets.DOCKERHUB_USERNAME }} - DOCKERHUB_TOKEN: ${{ secrets.DOCKERHUB_TOKEN }} - run: | - echo "username=[$DOCKERHUB_USERNAME]" - echo "token_length=${#DOCKERHUB_TOKEN}" - - name: Log in to Docker Hub if: github.event_name != 'pull_request' env: diff --git a/.github/workflows/release-publish.yml b/.github/workflows/release-publish.yml index 94f7d0c6..f9a0dd7e 100644 --- a/.github/workflows/release-publish.yml +++ b/.github/workflows/release-publish.yml @@ -1,5 +1,59 @@ name: Release Publish +on: + push: + branches: + - main + +env: + IMAGE_NAME: bankofai/docs + +jobs: + build-and-publish: + runs-on: ubuntu-latest + permissions: + id-token: write + contents: read + + steps: + - name: Checkout repository + uses: actions/checkout@v4 + + - name: Set up Docker Buildx + uses: docker/setup-buildx-action@v3 + + - name: Log in to Docker Hub + env: + DOCKERHUB_USERNAME: ${{ secrets.DOCKERHUB_USERNAME }} + DOCKERHUB_TOKEN: ${{ secrets.DOCKERHUB_TOKEN }} + run: | + echo "$DOCKERHUB_TOKEN" | docker login -u "$DOCKERHUB_USERNAME" --password-stdin + + - name: Get version from package.json + id: version + run: echo "VERSION=$(node -p "require('./package.json').version")" >> $GITHUB_OUTPUT + + - name: Set Docker tags + id: meta + run: | + echo "tags<> $GITHUB_OUTPUT + echo "${{ env.IMAGE_NAME }}:${{ steps.version.outputs.VERSION }}" >> $GITHUB_OUTPUT + echo "EOF" >> $GITHUB_OUTPUT + + - name: Build and push Docker image + uses: docker/build-push-action@v6 + with: + context: . + push: true + tags: ${{ steps.meta.outputs.tags }} + labels: ${{ steps.meta.outputs.labels }} + build-args: | + APP_ENV=production + cache-from: type=gha + cache-to: type=gha,mode=max + platforms: linux/amd64 +name: Release Publish + on: push: branches: diff --git a/docs/llmservice/models/deepseek-v3.2.md b/docs/llmservice/models/deepseek-v3.2.md index 9fdfb098..ce159e38 100644 --- a/docs/llmservice/models/deepseek-v3.2.md +++ b/docs/llmservice/models/deepseek-v3.2.md @@ -1,4 +1,6 @@ -# DeepSeek-V3.2 Model Overview +# DeepSeek-V3.2 + +## Overview **DeepSeek-V3.2** is a **reasoning-first** large language model launched by DeepSeek AI, specifically built for **Agentic scenarios**. By implementing the innovative **DeepSeek Sparse Attention (DSA)** mechanism, it achieves exceptional computational efficiency while demonstrating superior performance in complex tasks and autonomous agent capabilities. diff --git a/i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/deepseek-v3.2.md b/i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/deepseek-v3.2.md index ac5f6365..95fb05b2 100644 --- a/i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/deepseek-v3.2.md +++ b/i18n/zh-Hans/docusaurus-plugin-content-docs/current/llmservice/models/deepseek-v3.2.md @@ -1,4 +1,6 @@ -# DeepSeek-V3.2 概览 +# DeepSeek-V3.2 + +## 概述 **DeepSeek-V3.2** 是由 DeepSeek AI 推出的**推理导向型 (Reasoning-first)** 大语言模型，专为 **智能体 (Agent)** 场景量身打造。该模型通过创新的 **DSA (DeepSeek Sparse Attention)** 稀疏注意力机制极大优化了计算效率，在复杂任务处理与自主代理能力上表现卓越。 diff --git a/i18n/zh-Hans/docusaurus-plugin-content-docs/current/sidebars.js b/i18n/zh-Hans/docusaurus-plugin-content-docs/current/sidebars.js index c154bf8f..be6ba33e 100644 --- a/i18n/zh-Hans/docusaurus-plugin-content-docs/current/sidebars.js +++ b/i18n/zh-Hans/docusaurus-plugin-content-docs/current/sidebars.js @@ -176,6 +176,7 @@ const sidebars = { 'llmservice/models/claude-opus-4-6', 'llmservice/models/claude-sonnet-4-5', 'llmservice/models/claude-sonnet-4-6', + 'llmservice/models/deepseek-v3.2', 'llmservice/models/gemini-3-1-pro', 'llmservice/models/gemini-3-flash', 'llmservice/models/glm-5', diff --git a/package.json b/package.json index 300615c9..6a20518a 100644 --- a/package.json +++ b/package.json @@ -1,6 +1,6 @@ { "name": "@x402-tron/docs", - "version": "1.2.3", + "version": "1.2.5", "description": "x402-tron documentation", "license": "MIT", "scripts": { diff --git a/sidebars.js b/sidebars.js index 8c6b3e56..6d664cc0 100644 --- a/sidebars.js +++ b/sidebars.js @@ -172,6 +172,7 @@ const sidebars = { 'llmservice/models/claude-opus-4-6', 'llmservice/models/claude-sonnet-4-5', 'llmservice/models/claude-sonnet-4-6', + 'llmservice/models/deepseek-v3.2', 'llmservice/models/gemini-3-1-pro', 'llmservice/models/gemini-3-flash', 'llmservice/models/glm-5', From eaff73df58713e45af891e8641428227af605e34 Mon Sep 17 00:00:00 2001 From: jerryji-prog Date: Mon, 30 Mar 2026 16:30:09 +0800 Subject: [PATCH 5/5] delete Duplicate content --- .github/workflows/release-publish.yml | 54 --------------------------- 1 file changed, 54 deletions(-) diff --git a/.github/workflows/release-publish.yml b/.github/workflows/release-publish.yml index f9a0dd7e..94f7d0c6 100644 --- a/.github/workflows/release-publish.yml +++ b/.github/workflows/release-publish.yml @@ -1,59 +1,5 @@ name: Release Publish -on: - push: - branches: - - main - -env: - IMAGE_NAME: bankofai/docs - -jobs: - build-and-publish: - runs-on: ubuntu-latest - permissions: - id-token: write - contents: read - - steps: - - name: Checkout repository - uses: actions/checkout@v4 - - - name: Set up Docker Buildx - uses: docker/setup-buildx-action@v3 - - - name: Log in to Docker Hub - env: - DOCKERHUB_USERNAME: ${{ secrets.DOCKERHUB_USERNAME }} - DOCKERHUB_TOKEN: ${{ secrets.DOCKERHUB_TOKEN }} - run: | - echo "$DOCKERHUB_TOKEN" | docker login -u "$DOCKERHUB_USERNAME" --password-stdin - - - name: Get version from package.json - id: version - run: echo "VERSION=$(node -p "require('./package.json').version")" >> $GITHUB_OUTPUT - - - name: Set Docker tags - id: meta - run: | - echo "tags<> $GITHUB_OUTPUT - echo "${{ env.IMAGE_NAME }}:${{ steps.version.outputs.VERSION }}" >> $GITHUB_OUTPUT - echo "EOF" >> $GITHUB_OUTPUT - - - name: Build and push Docker image - uses: docker/build-push-action@v6 - with: - context: . - push: true - tags: ${{ steps.meta.outputs.tags }} - labels: ${{ steps.meta.outputs.labels }} - build-args: | - APP_ENV=production - cache-from: type=gha - cache-to: type=gha,mode=max - platforms: linux/amd64 -name: Release Publish - on: push: branches: