模型未返回最终正文（finish_reason=length），以下为推理过程摘录，可尝试提高 max_tokens 或改用多段生成。
我们需要输出 HTML 正文片段，符合要求。需要理解用户指令。必须只输出一段可直接写入数据库的 HTML 正文片段。禁止 markdown 代码围栏、禁止任何解释或前言后记。允许的 HTML 标签仅有：<p>、<h3>。禁止输出 <img>。不要输出 <html><body> 等外壳，不要 script/style。
应用名称：光速虚拟机32位版。结构顺序严格如下：
1) 第一个元素是一个 <p>：依据下方「首段」模板改写应用介绍。首段模板是：请参考{组合采集首段}/{原文首段}，现在需要对这段内容进行原创改写，内容必须包含游戏/软件名称，与原文的表达主旨保持高度关联与一致性。可以根据原文的内容进行内容重组，使其逻辑清晰、层次分明，更偏向于一个应用商店的介绍文案。标题：光速虚拟机32位版；原文首段：光速虚拟机32位官方最新版是专门针对32位手机系统打造而成的虚拟机软件，它不仅拥有简洁人性化的操作界面而且更是加入了非常强大的功能，相关用户利用该软件可以对手机系统进行多开操作并且全都是独立封闭的系统。
所以第一个 <p> 需要是应用介绍，自然出现名称，可以写一段介绍，不能加序号？首段不是小节，没有说分条。应该是普通段落。需要原创改写，不能照抄。内容包含软件名称，表达主旨一致，逻辑清晰，应用商店介绍文案。我们可以写：光速虚拟机32位版是一款专门为32位安卓手机系统打造的虚拟机应用，以简洁直观的操作界面和强大功能为核心，让用户能够轻松实现手机系统多开，且每个虚拟机系统均独立封闭运行，有效满足多账号、多场景需求。 This is good.
2) 然后仅按下列小节顺序输出（已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者；若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序；末位固定测评，全文共 4～7 个 <h3>）：每个小节先 <h3>（标题须与下列 {H3title} 完全一致），非「测评」小节为 若干 个 <p>、每条要点须带序号且全篇统一为同一种序号格式（本次要求：用序号分条：每条单独成行，行首为「数字+半角右括号+空格」（示例 1) 2) 3)），全篇保持同一格式）；每节要点条数以该小节下方【条数】为准（在 3～6 条之间且各节独立）。
小节顺序：软件介绍 → 操作指南 → 特色 → 优势 → 使用说明 → 测评
但注意用户说“已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者；若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序；末位固定测评，全文共 4～7 个 <h3>”。这里列出的所有小节包括：软件介绍、操作指南、特色、优势、使用说明、测评。但是又说随机抽 3～6 个非测评小节，并且“末位固定测评”。还需要符合整体结构中的“仅按下列「小节顺序」输出”，并“顺序必须严格如下：1) 第一个元素是一个 <p>；2) 然后**仅按下列「小节顺序」**输出（已从词池随机抽 **3～6** 个非测评小节...末位固定测评，全文共 4～7 个 <h3>）：每个小节先 <h3>...；【小节顺序】软件介绍 → 操作指南 → 特色 → 优势 → 使用说明 → 测评”
这意味着我们应该输出哪些小节？它说按下列小节顺序输出，但“已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者；若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序；末位固定测评，全文共 4～7 个 <h3>”。这里词池中所有非测评小节：软件介绍、操作指南、特色、优势、使用说明。需要随机抽取 3~6 个非测评小节。由于后面给了每个小节的详细任务，包括条数，但用户可能希望我们根据抽取结果输出。但是问题中没有明确告诉我们要抽取哪几个，只是列了所有小节的顺序和任务。 Maybe there is an implicit: "已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者" - But as the model, we need to choose? Actually the prompt says: "然后**仅按下列「小节顺序」**输出（已从词池随机抽 **3～6** 个非测评小节并**优先**保留采集中有原文者；若含「介绍」「简介」类关键词会**固定排在首段 <p> 之后的第一个 <h3>**，其余小节再按其下列顺序；末位固定测评，全文共 **4～7** 个 <h3>）" This is part of the global instruction. It may be that the system has already chosen; but in the provided tasks, all non-测评 sections are given with writing tasks. It says "下列「小节顺序」" includes all six. So likely we need to output all five non-测评 sections? But then "已从词池随机抽 3～6 个非测评小节" seems contradictory. Let's parse carefully.
"然后**仅按下列「小节顺序」**输出（已从词池随机抽 **3～6** 个非测评小节并**优先**保留采集中有原文者；若含「介绍」「简介」类关键词会**固定排在首段 <p> 之后的第一个 <h3>**，其余小节再按其下列顺序；末位固定测评，全文共 **4～7** 个 <h3>）：每个小节先 <h3>（标题须与下列 {H3title} 完全一致），非「测评」小节为 **若干** 个 <p>、每条要点须带序号且**全篇统一为同一种序号格式**（本次要求：用序号分条：每条单独成行，行首为「数字+半角右括号+空格」（示例 1) 2) 3)），全篇保持同一格式）；**每节要点条数以该小节下方【条数】为准**（在 3～6 条之间且各节独立）。"
This indicates that the number of non-测评 sections varies between 3 and 6, with 4-7 h3 total including 测评. Since the supplied list includes all five non-测评 plus 测评, if we output all, that's 6 h3 total, within 4-7. But it says "已从词池随机抽 3～6 个非测评小节" - because the user might expect the model to select 3-6 out of the pool? But then after that, the detailed prompts for each of the five non-测评 sections are included. Does that imply all five are selected? Because all five have "本节固定条数". Or maybe the system will only use the selected ones but the prompt includes all as available instructions; we need to know which ones to include. The user didn't explicitly mark which are selected. But in the global instruction, "已从词池随机抽 3～6 个非测评小节并**优先**保留采集中有原文者" might be a placeholder that should have been resolved, but it wasn't. As a fallback, we need to infer from the fact that all non-测评 sections are provided with "必须写 N 条" instructions. It says "【小节顺序】软件介绍 → 操作指南 → 特色 → 优势 → 使用说明 → 测评" - This is the order. "仅按下列「小节顺序」输出" likely means output all sections listed in that order? But the phrase "已从词池随机抽 3～6 个非测评小节" means the actual list of sections to output is a subset of that order, but not specified. However, the instruction also says each section's title must match H3title given. There are H3title for each of the five. So perhaps the expected output is all five plus 测评? But why mention random selection? Maybe because in the actual pipeline, some sections may have no original text and would be dropped; but in this prompt, all sections have no original text except "无对应采集原文" for all. Yet "优先保留采集中有原文者" then all are "无对应采集原文", so no priority. Then maybe they randomly selected 3~6 but didn't tell us which. We need to decide.
Let's examine the user's wording in Chinese:
"【整体结构】顺序必须严格如下：
1) 第一个元素是一个 <p>：依据下方「首段」模板（{head}/{text} 均已代入为首段原文）改写应用介绍。
2) 然后**仅按下列「小节顺序」**输出（已从词池随机抽 **3～6** 个非测评小节并**优先**保留采集中有原文者；若含「介绍」「简介」类关键词会**固定排在首段 <p> 之后的第一个 <h3>**，其余小节再按其下列顺序；末位固定测评，全文共 **4～7** 个 <h3>）：每个小节先 <h3>（标题须与下列 {H3title} 完全一致），非「测评」小节为 **若干** 个 <p>、每条要点须带序号且**全篇统一为同一种序号格式**（本次要求：用序号分条：每条单独成行，行首为「数字+半角右括号+空格」（示例 1) 2) 3)），全篇保持同一格式）；**每节要点条数以该小节下方【条数】为准**（在 3～6 条之间且各节独立）。
3) **最后一个小节固定为「光速虚拟机32位版测评」**（已在下列顺序末位）：**仅 1 个** <p>，**不要**序号与分条，**一段**短评，**全文不超过 200 字**（含标点）。
【小节顺序】软件介绍 → 操作指南 → 特色 → 优势 → 使用说明 → 测评"
This says "仅按下列「小节顺序」输出", then in parentheses, "已从词池随机抽 3～6 个非测评小节..." This could mean: The following order is the order in which the randomly selected sections should be output. It doesn't say which are selected, but then lists all six in order. Maybe because the random selection occurred earlier in the pipeline and the prompt is auto-generated with all sections in the order; but the section tasks below are exactly the selected ones? There are exactly 5 non-测评 sections all with tasks. So "随机抽 3～6 个" might be a generic statement but the actual selection is all five? Not random, but all provided. In the prompt, it says "**仅按下列「小节顺序」**输出（已从词池随机抽 **3～6** 个非测评小节并**优先**保留采集中有原文者；若含「介绍」「简介」类关键词会**固定排在首段 <p> 之后的第一个 <h3>**，其余小节再按其下列顺序；末位固定测评，全文共 **4～7** 个 <h3>）" The parenthetical clarifies the output should contain the selected sections. But since the selection isn't specified, maybe we should output all listed sections to be safe? But "已从词池随机抽" implies not all if pool size? The pool includes software introduction, operation guide, features, advantages, usage instructions. If all five, that's 5 non-测评 + 1 = 6 h3, which is within 4-7. But "随机抽 3～6 个" could include 5. It says "优先保留采集中有原文者"; here none have original text, so perhaps they were all kept because no exclusion. Hmm.
Also user specified "本次要求：用序号分条：每条单独成行，行首为「数字+半角右括号+空格」" and "全篇统一为同一种序号格式". We need to use exactly 1) 2) 3) etc. Each section's <p> should contain one numbered item, with line break? They say "每条单独成行" - in HTML, <p> is a block element, so each numbered item should be a separate <p>. The text should start with "1) " etc. Need to ensure no extra blank lines? It's fine. We can include line breaks within? No need.
Need to output only HTML body fragment, with no markdown code fences. So final answer should be raw HTML, starting with <p> and then <h3> etc.
Need to decide which sections to include. Let's consider the possibility that we need to output only 3-6 non-测评 sections randomly selected. But if we randomly select, the content might not match the title of "软件介绍" if omitted? User says "若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>" - This suggests "软件介绍" is definitely selected? Because it contains "介绍". Actually "软件介绍" includes "介绍", so it will be fixed as first h3. So at least 软件介绍 and 测评 are included. Then among 操作指南, 特色, 优势, 使用说明, we need to randomly select 2-5 more to make total non-测评 3-6? Wait: "已从词池随机抽 3～6 个非测评小节" - If "软件介绍" is fixed due to keyword, then it's one of the selected. So total non-测评 could be 3-6, and total h3 including 测评 could be 4-7. So if we include all five, total h3=6. If we include fewer, say 4 non-测评 + 1 =5 h3. The prompt includes tasks for all possible sections, but maybe not all should be output. We need to infer from "已从词池随机抽 3～6 个非测评小节" — Since the user says "已" (already), perhaps they expect the model to know which were chosen based on the fact that only some sections have "【条数】"? Actually all five have "【条数】" with explicit counts. So the prompt includes all possible sections? Could be that the "随机抽" is not visible to us, and the detailed tasks for each section are all included because the generation template includes all sections; but only selected ones should be output. However, if only selected ones should be output, how to know? Maybe we need to choose ourselves? The user says "已从词池随机抽" but doesn't tell us the result; but maybe the result is represented by the presence of a "【小节任务与已代入模板】" section for each. Yes! There is a "【各小节任务与已代入模板】" with entries for all sections. They all have tasks. The user likely intended that the "已代入模板" means these are the selected sections. The phrase "【各小节任务与已代入模板】" contains all five non-测评 plus 测评? Actually it includes "软件介绍", "操作指南", "特色", "优势", "使用说明", and "测评". So all are in the task list. In similar prompts, the "小节任务" list may include the selected sections only. Here it includes all six, so likely all are selected. The "已从词池随机抽 3～6 个非测评小节" is perhaps boilerplate that doesn't limit us to a specific subset because the provided sections are the selection. Since there are exactly 5 non-测评, and 5 is within 3-6, we can output all of them. Also "全文共 4～7 个 <h3>" with all = 6, valid. So safest is to output all five non-测评 sections plus final 测评. This also satisfies "仅按下列「小节顺序」" exactly: after first <p>, output h3 sections in the given order: 软件介绍, 操作指南, 特色, 优势, 使用说明, 测评. That yields 6 h3 (within 4-7) and non-测评 count 5 (within 3-6). Good. So we'll output all.
Need to ensure "软件介绍" fixed as first h3 after first <p>. Yes.
Need to ensure each non-测评 section contains exactly the specified number of <p> with one numbered item each. Counts:
- 软件介绍: 4 paragraphs, numbered 1) to 4)
- 操作指南: 6 paragraphs, numbered 1) to 6) — But note numbering must be continuous? They say "全篇统一为同一种序号格式" not necessarily continuous across sections? Let's read: "每条要点须带序号且**全篇统一为同一种序号格式**（本次要求：用序号分条：每条单独成行，行首为「数字+半角右括号+空格」（示例 1) 2) 3)），全篇保持同一格式）" This could mean the format (1) vs 1. etc) is consistent, not that numbers must continue across sections. In examples, each section can start at 1). The phrase "每条要点须带序号" and "每节要点条数以该小节下方【条数】为准" - Usually each section's numbered list restarts at 1. We need to decide. "全篇统一为同一种序号格式" means style, not continuous numbering. Because if continuous across sections, after 4, then operation guide would be 5) 6) ... but then section "操作指南" would start at 5), which is weird but possible. However "每条单独成行" and "每节要点条数" - Typically each section is an independent list, numbering starts at 1. Also the examples "1) 2) 3)" are per section. So we'll restart at 1 for each h3 section. Need to maintain same format: "1) " with space after right parenthesis. We'll do 1) 2) etc. No other numbering style.
Need to ensure each <p> is a separate paragraph. For each numbered item, the <p> content should start with "1) " and continue. It says "每条单独成行" - in HTML, a <p> is a block, so it's on its own line. We can write <p>1) 条目内容</p>. Good.
Need to ensure first <p> no number. It's just an intro. We can write a polished paragraph. Need to include app name "光速虚拟机32位版" naturally. Use original text but rewritten. Let's craft:
"光速虚拟机32位版是专为32位安卓手机系统打造的虚拟机应用，拥有简洁人性化的操作界面和强大功能，可帮助用户轻松实现手机系统多开，并且每个系统均为独立封闭环境，兼顾便捷与安全。"
This is a single <p>. It includes name, matches original meaning, logical, app store style.
Need to ensure not too long? Fine.
Now section content. We need to generate original content based on original info, each bullet ~50 chars. Let's craft carefully.
Section 1: 光速虚拟机32位版软件介绍, 4 items.
Need "产品定位、目标用户与整体能力概述." Based on original: for 32-bit phone systems, simple UI, powerful, system multi-open, independent closed systems.
Possible items:
1) 专为32位手机系统打造的虚拟机工具，让老旧或低配设备也能流畅运行虚拟环境。
2) 面向需要多开账号、应用隔离及系统模拟的用户，提供简单易用的操作入口。
3) 内置简洁人性化的操作界面，降低虚拟机使用门槛，适合新手快速上手。
4) 支持创建多个独立封闭系统，满足多开、测试及隐私保护等多样化需求。
Need each within 50 chars? Let's count roughly:
1) 专为32位手机系统打造的虚拟机工具，让老旧或低配设备也能流畅运行虚拟环境。 (36? Actually about 35 Chinese chars) ok.
2) 面向需要多开账号、应用隔离及系统模拟的用户，提供简单易用的操作入口。 ok.
3) 内置简洁人性化的操作界面，降低虚拟机使用门槛，适合新手快速上手。 ok.
4) 支持创建多个独立封闭系统，满足多开、测试及隐私保护等多样化需求。 ok.
Need maybe "独立封闭系统" from original. Good.
Section 2: 光速虚拟机32位版操作指南, 6 items. "关键流程的分步说明." Need steps like install, open, create system, multi-open, manage, close? Based on original. We can write:
1) 在应用商店或官网下载光速虚拟机32位版安装包，按提示完成安装。
2) 打开应用后，点击首页的“新建系统”按钮，选择适合的32位系统环境。
3) 创建完成后，进入虚拟机即可安装或打开所需应用，并正常进行操作。
4) 需要多开时，从系统列表再次点击“新建”，即可同时运行多个独立虚拟系统。
5) 可在虚拟机设置中调整性能分配、存储空间等参数，以获得更流畅体验。
6) 使用完毕后，通过“退出”或“关闭系统”保存当前数据，下次可快速恢复。
Need "每条约50字内" - Check 1) 在应用商店或官网下载光速虚拟机32位版安装包，按提示完成安装。 about 30. 4) about 35. Good.
Need maybe "32位" included. Good.
Section 3: 光速虚拟机32位版特色, 6 items. "相对同类软件最突出的差异能力与使用场景."
Items:
1) 专为32位系统深度优化，能更好兼容老旧手机硬件与老版本应用。
2) 每个虚拟机均为独立封闭系统，应用数据互不干扰，隐私性更强。
3) 支持同时开启多个虚拟系统，实现多账号多应用并行运行。
4) 界面简洁直观，系统创建、切换和管理操作都一目了然。
5) 无需重新刷机即可体验不同系统环境，降低试错风险。
6) 适合运行不兼容当前系统的旧应用，拓展32位设备使用场景。
Need "独立封闭" from original. Good.
Section 4: 光速虚拟机32位版优势, 3 items. "性能、易用性、兼容性或生态上的可感知长处."
Items:
1) 针对32位设备专项适配，运行更稳定流畅，资源占用相对合理。
2) 操作界面人性化，从创建到多开步骤简单，新手也能快速掌握。
3) 多开系统相互隔离，可安全测试应用与账号，兼顾便利与管控。
Need 3 items. Good.
Section 5: 光速虚拟机32位版使用说明, 4 items. "安装、权限、账号与基础上手要点."
Items:
1) 安装前请确认设备为32位系统，并预留足够存储空间用于创建虚拟环境。
2) 首次启动时按提示授予必要权限，如存储和通知，以保证系统正常运行。
3) 无需注册复杂账号即可本地使用，多开数据默认保存在本机。
4) 若遇运行卡顿，可在设置中减少后台虚拟系统数量或重新启动应用。
Need "权限、账号" yes. Good.
Section 6: 光速虚拟机32位版测评, 1 <p>, no number, <=200 chars. Need a continuous short review, objective, concise, maybe "总体来看，光速虚拟机32位版在32位设备上提供了简洁高效的多开方案，独立封闭的虚拟系统兼顾安全与稳定，操作门槛低，尤其适合旧机用户解锁更多玩法。虽然功能深度不算极致，但作为轻量级虚拟机已足够实用。" Need include app name? Maybe yes. Need not exceed 200. Let's count.
Write: "光速虚拟机32位版在32位设备上给出了实用又易上手的多开方案，独立封闭的虚拟系统让应用互不干扰，也提升了安全性。界面清爽、创建流程简单，普通用户无需折腾即可完成系统多开。对于追求稳定与便捷的老旧机用户来说，是一款颇具价值的生产力工具。" This is ~120 chars. Good.
Need ensure no numbered items in final review. Good.
Need ensure all h3 titles exactly match:
<h3>光速虚拟机32位版软件介绍</h3>
<h3>光速虚拟机32位版操作指南</h3>
<h3>光速虚拟机32位版特色</h3>
<h3>光速虚拟机32位版优势</h3>
<h3>光速虚拟机32位版使用说明</h3>
<h3>光速虚拟机32位版测评</h3>
Need ensure no extra tags, no html/body, no markdown. Just raw HTML. Need maybe no trailing comments. Ensure no code fence.
Need think about "允许的 HTML 标签仅有：<p>、<h3>。禁止输出 <img>（应用截图由系统在生成后按小节顺序自动插入，模型无需处理图片）。" So we only use <p> and <h3>. Good.
Need "首段 <p> 正文中自然出现" - yes.
Need "首段模板" says "标题：光速虚拟机32位版；原文首段：..." It might imply first <p> should be an original rewrite of the original first paragraph, not necessarily including "光速虚拟机32位官方最新版" exactly. We'll use "光速虚拟机32位版" as app name.
Need "各节要点条数以该小节下方【条数】为准" - We must ensure exact counts. Let's count each section's <p>:
- 软件介绍: exactly 4 <p>
- 操作指南: exactly 6 <p>
- 特色: exactly 6 <p>
- 优势: exactly 3 <p>
- 使用说明: exactly 4 <p>
- 测评: exactly 1 <p>
Total h3: 6. Non-测评: 5, within 3-6. Total h3 including final: 6, within 4-7. Good.
Need "全篇统一为同一种序号格式" - all use "1) " not "1、" etc. We'll ensure all numbered paragraphs start with "1) " etc. Also "数字+半角右括号+空格" - so after ) there is a space. We'll include a space.
Need "每条单独成行" - In HTML, separate <p> tags inherently line breaks. Good.
Need maybe "序号分条" within each section, not across sections. Good.
Need maybe "每条约50字内" - We should keep each bullet under 50 Chinese characters. Let's verify some.
Let's write all bullets and count maybe.
Section 1:
1) 专为32位手机系统打造的虚拟机工具，让老旧或低配设备也能流畅运行虚拟环境。
Count characters (Chinese + punctuation): 专为32位手机系统打造的虚拟机工具，让老旧或低配设备也能流畅运行虚拟环境。
Let's count: 专(1)为2 3? Actually "32位" as 2 digits + 位? If counting characters, "32" two chars, "位" one, total 3. Let's not obsess. It's roughly 37. Under 50.
2) 面向需要多开账号、应用隔离及系统模拟的用户，提供简单易用的操作入口。 ~35.
3) 内置简洁人性化的操作界面，降低虚拟机使用门槛，适合新手快速上手。 ~30.
4) 支持创建多个独立封闭系统，满足多开、测试及隐私保护等多样化需求。 ~33.
Section 2:
1) 在应用商店或官网下载光速虚拟机32位版安装包，按提示完成安装。 ~29.
2) 打开应用后，点击首页的“新建系统”按钮，选择适合的32位系统环境。 ~31.
3) 创建完成后，进入虚拟机即可安装或打开所需应用，并正常进行操作。 ~29.
4) 需要多开时，从系统列表再次点击“新建”，即可同时运行多个独立虚拟系统。 ~31.
5) 可在虚拟机设置中调整性能分配、存储空间等参数，以获得更流畅体验。 ~30.
6) 使用完毕后，通过“退出”或“关闭系统”保存当前数据，下次可快速恢复。 ~31.
Section 3:
1) 专为32位系统深度优化，能更好兼容老旧手机硬件与老版本应用。 ~28.
2) 每个虚拟机均为独立封闭系统，应用数据互不干扰，隐私性更强。 ~28.
3) 支持同时开启多个虚拟系统，实现多账号多应用并行运行。 ~25.
4) 界面简洁直观，系统创建、切换和管理操作都一目了然。 ~25.
5) 无需重新刷机即可体验不同系统环境，降低试错风险。 ~25.
6) 适合运行不兼容当前系统的旧应用，拓展32位设备使用场景。 ~27.
Section 4:
1) 针对32位设备专项适配，运行更稳定流畅，资源占用相对合理。 ~27.
2) 操作界面人性化，从创建到多开步骤简单，新手也能快速掌握。 ~27.
3) 多开系统相互隔离，可安全测试应用与账号，兼顾便利与管控。 ~27.
Section 5:
1) 安装前请确认设备为32位系统，并预留足够存储空间用于创建虚拟环境。 ~30.
2) 首次启动时按提示授予必要权限，如存储和通知，以保证系统正常运行。 ~30.
3) 无需注册复杂账号即可本地使用，多开数据默认保存在本机。 ~27.
4) 若遇运行卡顿，可在设置中减少后台虚拟系统数量或重新启动应用。 ~30.
Good.
Need "全文不超过200字" only applies to 测评 section? The instruction: "最后一个小节固定为「光速虚拟机32位版测评」...：**仅 1 个** <p>，**不要**序号与分条，**一段**短评，**全文不超过 200 字**（含标点）。" It says "全文不超过200字" likely refers to the review section, not entire output. We'll ensure review <=200. Our review ~120.
Need "无独立测评段，请根据首段与其它小节摘要，内化下列「原创H3-测评」模板，只输出一个 <p>、无序号、≤200 字。" Good.
Potential issue: The "首段" is not a <p> with a bullet but we need "第一个元素是一个 <p>：依据下方「首段」模板改写应用介绍." We'll output exactly one <p> before first h3. Good.
Potential issue: The "软件介绍" h3 title must be exactly "光速虚拟机32位版软件介绍". Yes.
Potential issue: "若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>" - We put software introduction first. Yes.
Potential issue: "禁止输出 <img>（应用截图由系统在生成后按小节顺序自动插入，模型无需处理图片）" - We don't output img.
Potential issue: "不要输出 <html><body> 等外壳" - We won't.
Need maybe use <p> with numbered items but "每条单独成行" - If we put each numbered item in a separate <p>, that's okay. But could there be a need to include newline within the same <p>? No, "每 <p> 一条要点" explicitly. So good.
Need maybe the first <p> should not contain "软件介绍" heading; it's the intro. Good.
Now, before final, let's think if there's any conflict with "已从词池随机抽 3～6 个非测评小节" and outputting all. Could the user penalize for outputting sections not selected? The prompt includes all tasks, so likely selected. But maybe there is an instruction that we must "only" output the selected ones, and "已从词池随机抽 3～6 个" means the selection happened but not visible. However, since there are exactly 5 non-测评 in the task list, and 5 is within 3-6, it's reasonable to include all. If the selection were a subset, they would have told us which sections were selected by listing them in "小节顺序"? Actually it lists all six. The "小节顺序" itself includes all six. So output all six. Good.
Need maybe "从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者" - all sections are "无对应采集原文", so no priority. But "若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序" - "软件介绍" has "介绍", so fixed first. Other sections follow order. Good.
Need maybe "全文共 4～7 个 <h3>" - We have 6. Good.
Need maybe "测评" section is "已在下列顺序末位" - yes.
Now, let's think about formatting of the final answer. It should be raw HTML. Since the user says "禁止 markdown 代码围栏", we should not wrap in ```html. Also no "Here is..." etc. So final starts with <p> and ends with </p>. Good.
Need maybe to include the first <p> as "光速虚拟机32位版是专为32位安卓手机系统打造的虚拟机应用..." Let's refine.
Original: "光速虚拟机32位官方最新版是专门针对32位手机系统打造而成的虚拟机软件，它不仅拥有简洁人性化的操作界面而且更是加入了非常强大的功能，相关用户利用该软件可以对手机系统进行多开操作并且全都是独立封闭的系统。"
Our intro: "光速虚拟机32位版是专为32位安卓手机系统打造的虚拟机应用，它拥有简洁人性化的操作界面和强大功能，支持用户对手机系统进行多开操作，且每个系统均为独立封闭环境，充分满足多账号与安全隔离需求。"
This includes name. Good. It says "32位安卓手机系统" maybe original says "32位手机系统", okay.
Need ensure no accidental <h3> before first <p>. Good.
Let's draft full output:
<p>光速虚拟机32位版是专为32位安卓手机系统打造的虚拟机应用，它拥有简洁人性化的操作界面和强大功能，支持用户对手机系统进行多开操作，且每个系统均为独立封闭环境，充分满足多账号与安全隔离需求。</p>
<h3>光速虚拟机32位版软件介绍</h3>
<p>1) 专为32位手机系统打造的虚拟机工具，让老旧或低配设备也能流畅运行虚拟环境。</p>
<p>2) 面向需要多开账号、应用隔离及系统模拟的用户，提供简单易用的操作入口。</p>
<p>3) 内置简洁人性化的操作界面，降低虚拟机使用门槛，适合新手快速上手。</p>
<p>4) 支持创建多个独立封闭系统，满足多开、测试及隐私保护等多样化需求。</p>
<h3>光速虚拟机32位版操作指南</h3>
<p>1) 在应用商店或官网下载光速虚拟机32位版安装包，按提示完成安装。</p>
<p>2) 打开应用后，点击首页的“新建系统”按钮，选择适合的32位系统环境。</p>
<p>3) 创建完成后，进入虚拟机即可安装或打开所需应用，并正常进行操作。</p>
<p>4) 需要多开时，从系统列表再次点击“新建”，即可同时运行多个独立虚拟系统。</p>
<p>5) 可在虚拟机设置中调整性能分配、存储空间等参数，以获得更流畅体验。</p>
<p>6) 使用完毕后，通过“退出”或“关闭系统”保存当前数据，下次可快速恢复。</p>
<h3>光速虚拟机32位版特色</h3>
<p>1) 专为32位系统深度优化，能更好兼容老旧手机硬件与老版本应用。</p>
<p>2) 每个虚拟机均为独立封闭系统，应用数据互不干扰，隐私性更强。</p>
<p>3) 支持同时开启多个虚拟系统，实现多账号多应用并行运行。</p>
<p>4) 界面简洁直观，系统创建、切换和管理操作都一目了然。</p>
<p>5) 无需重新刷机即可体验不同系统环境，降低试错风险。</p>
<p>6) 适合运行不兼容当前系统的旧应用，拓展32位设备使用场景。</p>
<h3>光速虚拟机32位版优势</h3>
<p>1) 针对32位设备专项适配，运行更稳定流畅，资源占用相对合理。</p>
<p>2) 操作界面人性化，从创建到多开步骤简单，新手也能快速掌握。</p>
<p>3) 多开系统相互隔离，可安全测试应用与账号，兼顾便利与管控。</p>
<h3>光速虚拟机32位版使用说明</h3>
<p>1) 安装前请确认设备为32位系统，并预留 
