Skip to content

Make Codex and Claude Code subscriptions go further: a local model router that gives routine agent work to economical models and keeps your strongest model for the hard parts.

fix(website): localize article visuals and simplify navigation (#45)

澄潭8cd7e68c2aac
diff --git a/apps/website/README.md b/apps/website/README.md
index 58d98bdd..82d98531 100644
--- a/apps/website/README.md
+++ b/apps/website/README.md
@@ -27,12 +27,14 @@ Chinese lives at `/`; English lives at `/en/`. Language is an explicit link, not
an IP-based redirect. All pages are static directory indexes; `/releases/*` is
never part of the website build and must remain a real object/404 boundary.
-## News and experiment articles
+## Articles and case studies
The repository-root `news/index.json` indexes bilingual articles in `news/`.
Both GitHub and the website use those same Markdown sources. Add an entry with a
-unique slug, date, title, description, and file for each language; the news lists,
+unique slug, date, title, description, and file for each language; the article lists,
language links and newest-article homepage entry follow the index automatically.
+An optional localized `shortTitle` keeps the homepage link concise; without one,
+it uses the full article title. Public article URLs remain under `/news/`.
Article images live in `news/assets/` and are copied to ignored `public/news-assets/`
at build time. Evidence and reproduction commands belong in the public
`experiments/` directory. Keep these directories self-contained.
diff --git a/apps/website/src/components/HomePage.astro b/apps/website/src/components/HomePage.astro
index c606d3c8..4fda8445 100644
--- a/apps/website/src/components/HomePage.astro
+++ b/apps/website/src/components/HomePage.astro
@@ -7,6 +7,7 @@ const en = language === 'en';
const t = (zh, english) => en ? english : zh;
const path = en ? '/en/' : '/';
const alternatePath = en ? '/' : '/en/';
+const featuredPost = posts[0];
const release = latestRelease('stable');
const preferredArtifact = release?.artifacts.find(item => item.kind === 'desktop' && item.architecture === 'arm64')
?? release?.artifacts.find(item => item.kind === 'desktop');
@@ -19,7 +20,11 @@ const releaseKinds = [preferredArtifact && t('macOS 桌面应用', 'macOS Deskto
path={path} alternatePath={alternatePath} fullWidth>
<section class="hero wrap">
<div>
- <div class="eyebrow section-label">{t('面向长程 Agent 的智能路由', 'Intelligent routing for long-horizon agents.')}</div>
+ {featuredPost ? <a class="hero-article" href={newsRoute(featuredPost.slug, language)} title={featuredPost.title[language]}>
+ <span class="hero-article-label">{t('最新文章', 'Latest article')}</span>
+ <span class="hero-article-copy">{featuredPost.shortTitle?.[language] ?? featuredPost.title[language]}</span>
+ <span class="hero-article-arrow" aria-hidden="true">→</span>
+ </a> : <div class="eyebrow section-label">{t('面向长程 Agent 的智能路由', 'Intelligent routing for long-horizon agents.')}</div>}
<h1>{en ? <><span>Intelligent routing</span><span>for long-horizon agents.</span></> : <><span>为长程 Agent 而生的</span><span>智能路由引擎。</span></>}</h1>
<p class="intro">{t('一段长任务,不必始终依赖同一个模型。按任务选择 Agent,再根据执行表现与后续工作的需要,决定继续使用还是切换模型。', 'A long task need not rely on one model from start to finish. Match the agent to the task, then choose the right model for each stage—guided by past performance and the work ahead.')}</p>
<a class="jev-callout" href="#how"><strong>Jev</strong><span>{t('官方决策扩展 · 一次调用,选择与评分', 'Official extension · One call to choose and assess')}</span><span aria-hidden="true">↗</span></a>
@@ -36,12 +41,6 @@ const releaseKinds = [preferredArtifact && t('macOS 桌面应用', 'macOS Deskto
</section>
<div class="trust-line wrap"><span><b>{t('更省', 'Efficient')}</b> · {t('按需投入算力', 'Compute where it counts')}</span><span><b>{t('更稳', 'Reliable')}</b> · {t('有边界地接力', 'Bounded failover')}</span><span><b>{t('更懂', 'Adaptive')}</b> · {t('用执行事实改进选择', 'Decisions grounded in execution')}</span></div>
- {posts[0] && <aside class="wrap news-feature" aria-label={t('最新文章', 'Latest article')}>
- <span class="news-feature-label">{t('最新实战', 'NEW CASE STUDY')}</span>
- <div><a href={newsRoute(posts[0].slug, language)}><strong>{posts[0].title[language]}</strong><span aria-hidden="true"> ↗</span></a><p>{posts[0].description[language]}</p></div>
- <a class="text-link" href={newsRoute('', language)}>{t('全部新闻', 'All news')} →</a>
- </aside>}
-
<section class="section wrap lifecycle-section" id="boundaries">
<div class="section-title"><div><div class="eyebrow section-label">{t('长任务不只做一次模型选择', 'A long task is not one model choice')}</div><h2>{t('阶段内保持稳定,', 'Stay stable within a stage.')}<br>{t('在自然边界重新选择。', 'Choose again at natural boundaries.')}</h2></div><p>{t('不必等用户再次提问。长会话压缩并重建上下文时,也能成为重新选模的自然机会;普通工具续轮保持当前模型。', 'A new user prompt is not the only boundary. When a long session compacts and rebuilds its context, HiRoute gets another natural chance to choose; ordinary tool continuations stay on the current model.')}</p></div>
<div class="timeline-legend"><span class="legend-economy">{t('经济模型', 'Economy model')}</span><span class="legend-primary">{t('主力模型', 'Primary model')}</span><span>{t('选择决策节点查看接续方式 · 机制示例', 'Select a decision point to inspect the handoff · Illustration')}</span></div>
diff --git a/apps/website/src/components/NewsPage.astro b/apps/website/src/components/NewsPage.astro
index 777e546e..1a6420f6 100644
--- a/apps/website/src/components/NewsPage.astro
+++ b/apps/website/src/components/NewsPage.astro
@@ -4,15 +4,15 @@ import { posts, newsRoute, renderNews } from '../lib/news.mjs';
const { language = 'zh', slug } = Astro.props;
const en = language === 'en';
const post = slug ? posts.find(p => p.slug === slug) : null;
-const title = post ? post.title[language] : (en ? 'News · HiRoute' : '新闻与实战 · HiRoute');
-const description = post ? post.description[language] : (en ? 'HiRoute experiments, engineering stories, and project news.' : 'HiRoute 的实验结果、工程实践与项目动态。');
+const title = post ? post.title[language] : (en ? 'Articles · HiRoute' : '文章 · HiRoute');
+const description = post ? post.description[language] : (en ? 'HiRoute case studies, engineering stories, and project updates.' : 'HiRoute 的实战案例、工程实践与项目动态。');
const body = post ? await renderNews(slug, language) : '';
---
<SiteLayout language={language} title={title} description={description}
path={newsRoute(slug, language)} alternatePath={newsRoute(slug, en ? 'zh' : 'en')}>
<div class="wrap news-shell">
{post ? <>
- <a class="text-link" href={newsRoute('', language)}>← {en ? 'All news' : '全部文章'}</a>
+ <a class="text-link" href={newsRoute('', language)}>← {en ? 'All articles' : '全部文章'}</a>
<article class="news-article">
<header><time datetime={post.date}>{post.date}</time><h1>{title}</h1></header>
<div class="news-prose" set:html={body} />
@@ -20,7 +20,7 @@ const body = post ? await renderNews(slug, language) : '';
</article>
</> : <>
<p class="eyebrow">{en ? 'From the HiRoute team' : '来自 HiRoute 团队'}</p>
- <h1>{en ? 'News & field notes' : '新闻与实战'}</h1>
+ <h1>{en ? 'Articles' : '文章'}</h1>
<p>{description}</p>
<div class="news-list">{posts.map(p => <article class="news-card"><time datetime={p.date}>{p.date}</time><h2><a href={newsRoute(p.slug, language)}>{p.title[language]}</a></h2><p>{p.description[language]}</p><a class="text-link" href={newsRoute(p.slug, language)}>{en ? 'Read the article' : '阅读全文'} →</a></article>)}</div>
</>}
diff --git a/apps/website/src/layouts/SiteLayout.astro b/apps/website/src/layouts/SiteLayout.astro
index 14a1c46a..e44796c0 100644
--- a/apps/website/src/layouts/SiteLayout.astro
+++ b/apps/website/src/layouts/SiteLayout.astro
@@ -11,11 +11,11 @@ const alternate = new URL(alternatePath, Astro.site);
const zhPath = english ? alternatePath : path;
const enPath = english ? path : alternatePath;
const nav = english ? [
- ['Product', '/en/#product'], ['How it works', '/en/#boundaries'],
- ['News', '/en/news/'], ['Docs', '/en/docs/'], ['Decision extensions', '/en/docs/decision-extensions/'], ['Download', '/en/download/'],
+ ['Product', '/en/#product'], ['Docs', '/en/docs/'],
+ ['Articles', '/en/news/'], ['Download', '/en/download/'],
] : [
- ['产品', '/#product'], ['如何工作', '/#boundaries'],
- ['新闻', '/news/'], ['文档', '/docs/'], ['决策扩展', '/docs/decision-extensions/'], ['下载', '/download/'],
+ ['产品', '/#product'], ['文档', '/docs/'],
+ ['文章', '/news/'], ['下载', '/download/'],
];
---
<!doctype html>
@@ -71,7 +71,7 @@ const nav = english ? [
</a>
<p>{english ? 'Intelligent routing for long-horizon agents.' : '面向长程 Agent 的智能路由引擎。'}</p>
<div class="footer-links">
- <a href={english ? '/en/news/' : '/news/'}>{english ? 'News' : '新闻'}</a>
+ <a href={english ? '/en/news/' : '/news/'}>{english ? 'Articles' : '文章'}</a>
<a href={english ? '/en/docs/' : '/docs/'}>{english ? 'Docs' : '文档'}</a>
<a href={english ? '/en/changelog/' : '/changelog/'}>{english ? 'Changelog' : '更新日志'}</a>
<a href={githubUrl} target="_blank" rel="noreferrer">GitHub ↗</a>
diff --git a/apps/website/src/styles/site.css b/apps/website/src/styles/site.css
index 9674665b..f738e8bb 100644
--- a/apps/website/src/styles/site.css
+++ b/apps/website/src/styles/site.css
@@ -349,7 +349,7 @@ html:lang(en) .section-title>p{flex:0 1 335px}
@media(max-width:760px){.engine-map{padding:20px}.engine-center{padding:11px}.engine-center img{width:45px;height:45px}.engine-center span{font-size:11px}.timeline-legend{gap:11px;font-size:10px}.timeline-grid{gap:12px}.decision-boundary small{flex-basis:100%}.decision-flow strong{font-size:15px}.ecosystem-groups{gap:25px}.agent-names{gap:8px}.agent-names span{font-size:12px}.faq-section summary{font-size:14px}}
@media(max-width:1000px){.history-bridge{grid-template-columns:1fr 2fr}.history-bridge a{grid-column:2}.boundary-detail{gap:24px}.specialization-flow{gap:12px}}
@media(max-width:760px){.boundary-detail{grid-template-columns:1fr;gap:22px}.boundary-outcome{padding-left:17px}.history-bridge{grid-template-columns:1fr;gap:14px;padding:20px}.history-bridge strong br{display:none}.history-bridge a{grid-column:1}.specialization{padding:22px}.specialization-heading h3{font-size:19px}.specialization-flow{grid-template-columns:1fr;gap:18px}.flow-arrow{transform:rotate(90deg)}.specialization-flow article{padding-top:15px}}
-@media(max-width:760px){html:lang(en) .nav{gap:14px}html:lang(en) .section-title>p{max-width:none}.language-switch a{padding:5px 8px}}
+@media(max-width:760px){html:lang(en) .nav{gap:14px}html:lang(en) .section-title>p{max-width:none}.language-switch a{padding:5px 8px}.site-header .language-switch{position:absolute;right:0;top:23px}}
@media(prefers-reduced-motion:reduce){html{scroll-behavior:auto}.decision-boundary{transition:none}}
/* Production website pages and documentation. */
@@ -362,7 +362,7 @@ html:lang(en) .section-title>p{flex:0 1 335px}
.page-hero h1{font-size:clamp(38px,5vw,58px);letter-spacing:-1.8px;margin:16px 0 19px}
.page-hero>p{max-width:720px;color:var(--muted);font-size:16px}
.content-section{padding:55px 0}
-.platform-install{display:grid;gap:24px}
+.platform-install{display:grid;grid-template-columns:minmax(0,1fr);gap:24px}
.platform-heading{display:flex;align-items:end;justify-content:space-between;gap:32px}
.platform-heading h2{font-size:28px;margin:15px 0 9px}
.platform-heading p{max-width:760px;color:var(--muted);font-size:14px}
@@ -480,13 +480,14 @@ html:lang(en) .section-title>p{flex:0 1 335px}
@media(max-width:760px){.hero{grid-template-columns:1fr;padding:40px 0;gap:35px}.hero h1{font-size:38px;letter-spacing:-1px}.journey-art{padding:24px 20px}.journey-stage h3{font-size:18px}.jev-heading,.jev-modes{grid-template-columns:1fr;gap:20px}.jev-diagram{overflow-x:auto;border-radius:16px}.jev-diagram img{min-width:740px}.jev-bottom p{font-size:17px}.section{padding:58px 0}.jev-modes article{padding:18px 0}.jev-heading h2{font-size:30px}}
/* News: canonical repository Markdown, readable on desktop and mobile. */
-.news-feature{display:flex;align-items:center;gap:24px;margin-top:32px;margin-bottom:24px;padding:24px;border:1px solid #d9def0;border-radius:18px;background:#eef1ff}
-.news-feature-label{flex-shrink:0;font-size:12px;font-weight:700;letter-spacing:.08em;color:#4255c8}
-.news-feature>div{flex:1;min-width:0}.news-feature p{margin:8px 0 0;font-size:14px;line-height:1.6;color:#535d72}.news-feature>a{flex-shrink:0}
+.hero-article{display:flex;align-items:center;gap:10px;width:fit-content;max-width:100%;padding:8px 10px;border:1px solid #dce1ef;border-radius:9px;background:#ffffffa6;color:#485cb5;font-size:12px;line-height:1.6}
+.hero-article:hover{border-color:#aeb8e8;background:#eef0fb}
+.hero-article-label{flex:none;padding:3px 7px;border-radius:4px;background:#e8ecfc;color:#4255b8;font-size:11px;font-weight:600}
+.hero-article-copy{min-width:0;text-wrap:pretty}.hero-article-arrow{flex:none;margin-left:auto}
.news-shell{max-width:1024px;padding-top:52px;padding-bottom:80px}.news-shell h1{font-size:clamp(28px,4vw,44px);line-height:1.3;letter-spacing:-.025em}
.news-article{max-width:860px;margin:24px auto}.news-article header{margin-bottom:32px}.news-shell time{font-size:14px;color:#606a7d}.news-prose{font-size:17px;line-height:1.85;overflow-wrap:anywhere}
.news-prose h2{font-size:27px;line-height:1.4;margin:48px 0 18px}.news-prose p,.news-prose ul,.news-prose ol{margin:20px 0}.news-prose ul,.news-prose ol{padding-left:26px}.news-prose li{margin:8px 0}
.news-prose img{display:block;max-width:100%;height:auto;border-radius:14px;margin:30px auto}.news-prose a{color:#4055ce;text-decoration:underline;text-underline-offset:3px}
.news-prose table{display:block;overflow-x:auto;border-collapse:collapse;width:100%;font-size:14px;line-height:1.6;margin:28px 0}.news-prose th,.news-prose td{border:1px solid #dce1eb;padding:12px;min-width:100px;text-align:left}.news-prose th{background:#eef1f8}
.news-prose pre{overflow-x:auto;background:#172034;color:#e6edf8;padding:20px;border-radius:12px;font-size:14px;line-height:1.65}.news-prose code{font-size:.9em}.news-list{display:grid;gap:24px;margin:36px 0}.news-card{border:1px solid #dce1eb;border-radius:18px;padding:28px;background:white}.news-card h2{font-size:25px;line-height:1.4}.news-card p{line-height:1.75;color:#535d72}
-@media(max-width:700px){.news-feature{align-items:flex-start;flex-direction:column;gap:12px}.news-shell{padding-top:32px}.news-prose{font-size:16px}.news-card{padding:20px}.news-prose h2{font-size:23px}.news-prose th,.news-prose td{padding:8px}}
+@media(max-width:700px){.news-shell{padding-top:32px}.news-prose{font-size:16px}.news-card{padding:20px}.news-prose h2{font-size:23px}.news-prose th,.news-prose td{padding:8px}}
diff --git a/news/2026-10-04-astra-qwen.en.md b/news/2026-10-04-astra-qwen.en.md
index 624bfa72..e2c37509 100644
--- a/news/2026-10-04-astra-qwen.en.md
+++ b/news/2026-10-04-astra-qwen.en.md
@@ -120,11 +120,11 @@ This case shows execution feedback influencing model selection throughout the ta
That execution feedback is also available for inspection. A session's **Model performance** view shows the actual model, competence score and assessment coverage for each execution stage. If progress looks insufficient, execution-evidence and assessment-feedback links lead back to the relevant activity. Stages that have not received a score remain visible too.
-[![Session model performance: inspect Qwen and Astra stage scores, assessment coverage and evidence links](assets/desktop-session-competence-zh.png)](assets/desktop-session-competence-zh.png)
+[![Session model performance: inspect Qwen and Astra stage scores, assessment coverage and evidence links](assets/desktop-session-competence-en.png)](assets/desktop-session-competence-en.png)
A routing plan's **Runtime performance** view provides a broader perspective. Select a plan version and time range to see each model's average stage competence, together with counts of scored and unrated stages. Open a model's stages to inspect the underlying execution and assess which combination fits your tasks. The average includes scored stages only; unrated stages do not enter the calculation.
-[![Routing-plan performance: inspect model averages and stage counts within one version and time range](assets/desktop-plan-competence-zh.png)](assets/desktop-plan-competence-zh.png)
+[![Routing-plan performance: inspect model averages and stage counts within one version and time range](assets/desktop-plan-competence-en.png)](assets/desktop-plan-competence-en.png)
## Enterprise gateway experience, lightweight local execution
diff --git a/news/ILLUSTRATIONS.md b/news/ILLUSTRATIONS.md
index 422fafb8..d014882c 100644
--- a/news/ILLUSTRATIONS.md
+++ b/news/ILLUSTRATIONS.md
@@ -98,9 +98,11 @@ as current product evidence or substitute a new score into the frozen case.
Two additional images explain the intended final observation interface:
-- [Session model performance](assets/desktop-session-competence-zh.png): actual models,
+- Session model performance ([Chinese](assets/desktop-session-competence-zh.png) /
+ [English](assets/desktop-session-competence-en.png)): actual models,
stage scores, assessment coverage and evidence links.
-- [Routing-plan performance](assets/desktop-plan-competence-zh.png): model averages
+- Routing-plan performance ([Chinese](assets/desktop-plan-competence-zh.png) /
+ [English](assets/desktop-plan-competence-en.png)): model averages
within one plan version and time range, with scored and unrated stage counts.
These images were generated with the built-in `image_gen` tool. They are product-design
@@ -121,8 +123,11 @@ Stage scores describe different stretches of work, not a controlled model rankin
or whole-delivery acceptance. Partial-assessment and historical-gap states remain
part of the underlying evidence; their notices belong in expanded evidence details
in this design. Removing them from the primary view does not alter the experiment
-records. Both language editions use the same two Chinese-interface images, with
-localized article text and captions.
+records. Each language edition uses matching localized interface images. The English
+versions translate the navigation, filters, labels, actions and explanatory text while
+preserving the source layout, model identities, numerical values, stage counts and
+unrated state. Their source-image hashes and exact localization prompts are retained
+in the same provenance record.
## Local architecture
diff --git a/news/README.md b/news/README.md
index 01657c84..344ef455 100644
--- a/news/README.md
+++ b/news/README.md
@@ -1,7 +1,7 @@
-# HiRoute News
+# HiRoute Articles
| Date | English | 简体中文 |
| --- | --- | --- |
| 2026-10-04 | [Smart routing: Astra × Qwen](2026-10-04-astra-qwen.en.md) | [智能路由省钱实战](2026-10-04-astra-qwen.zh-CN.md) |
-Articles here are the canonical source for the bilingual website News pages. Add a dated Markdown pair and an entry in `index.json` for future posts.
+Articles here are the canonical source for the bilingual website Articles pages. Add a dated Markdown pair and an entry in `index.json` for future posts. A localized `shortTitle` is optional for the compact homepage link.
diff --git a/news/assets/desktop-plan-competence-en.png b/news/assets/desktop-plan-competence-en.png
new file mode 100644
index 00000000..bf37899a
Binary files /dev/null and b/news/assets/desktop-plan-competence-en.png differ
diff --git a/news/assets/desktop-session-competence-en.png b/news/assets/desktop-session-competence-en.png
new file mode 100644
index 00000000..7c8b4589
Binary files /dev/null and b/news/assets/desktop-session-competence-en.png differ
diff --git a/news/competence-product-views.json b/news/competence-product-views.json
index 8ef9891a..02e51558 100644
--- a/news/competence-product-views.json
+++ b/news/competence-product-views.json
@@ -43,6 +43,26 @@
"Use case: precise-object-edit. Make a narrow edit to this HiRoute routing-plan interface illustration. Preserve layout, HiRoute logo, model names, 0.48 / 1 and 0.78 / 1 averages, score/sample counts, model cards, dropdowns, detail rows, action buttons, typography and sidebar.\nRemove the two amber-dot lines \"其中部分证据评分:1 段\" entirely from the model cards; do not put replacement text or badges there. Keep \"已评分 1 段 · 未评分 0 段\" on Qwen and \"已评分 1 段 · 未评分 1 段\" on Astra. Keep 查看阶段 actions and rebalance card whitespace neatly.\nIn the explanatory strip below the cards replace the entire sentence with exactly \"均值按已评分阶段计算,未评分阶段不计入。\"\nIn the lower 阶段明细 rows remove both amber \"部分证据\" badges, keeping all model names, execution ranges, numerical scores, the 未评分 state, and 执行证据/查看证据 actions unchanged. No alternative warning or completeness badge.\nPreserve the bottom explanatory sentence about not representing general model capability or task success rate, and preserve the footer \"界面效果示意 · 非原始实验截图\". No words 部分证据, 历史记录有缺口, 证据不完整, or 设计示意 anywhere in the image. Do not fabricate complete-evidence status. These details will remain accessible only when a user opens evidence in the future product implementation. No other changes; authentic flat HiRoute Desktop style, opaque background.",
"Use case: precise-object-edit. Edit this exact HiRoute routing-plan interface image with ONE change only. Remove the small bottom-right caption 界面效果示意 · 非原始实验截图, replacing the text pixels seamlessly with the surrounding app background. Preserve the entire app frame and original canvas dimensions; do not crop. Keep the bottom-left info sentence about stage performance and general capability unchanged. Preserve all UI panels, logos, model names, scores 0.48 / 1 and 0.78 / 1, sample counts, unrated state, dropdowns, evidence links, fonts, spacing and colors exactly as closely as possible. Do not add any replacement caption, watermark, disclaimer or status. This is a clean product-design asset for the intended final interface; provenance is stored separately. Opaque light background."
]
+ },
+ {
+ "path": "assets/desktop-session-competence-en.png",
+ "language": "en",
+ "sha256": "a91f87d8a76092ddb8ee685fe581a3b2331ed2e8ab6bcc6188734d2a5192a2db",
+ "source_image": "assets/desktop-session-competence-zh.png",
+ "source_sha256": "50c8c749e5837744ea7b548992c81d5242583ec5d55d777e2d4d70e0602d14be",
+ "edit_prompt": "Use case: text-localization.\nAsset type: English-language version of an existing HiRoute Desktop product UI illustration.\nInput image: the single referenced image is the edit target.\nPrimary request: Translate ALL Chinese UI text to English, preserving the existing full desktop frame, layout, sidebar widths, panels, icons, lavender accents, light background, typography hierarchy and data. This is localization of the same product-design asset, not a new scene. Preserve the original approximately 1.6:1 landscape ratio and complete app frame; use crisp high-resolution readable text. Never crop. Reflow or slightly reduce text size only where English needs room, without overlapping or truncating labels.\n\nExact English UI text to use:\nTop window title: \"Sessions\".\nLeft navigation: \"Home\", \"Models\", \"Smart Routing\", \"Agent\", \"Sessions\", \"Settings\". Keep the HiRoute logo and \"AGENT ROUTING\".\nBottom-left status: \"Service running\" and \"Runs locally\".\nSession search: \"Search conversations\".\nTabs: \"All\" and \"Request fallbacks\".\nSelected session title and large main title: \"HTTPX JSON Streaming\". Session subline remains \"Astra × Qwen\".\nUpper-right button: \"Run History\".\nMain section: \"Model Performance\".\nSection subtitle: \"Stage scores, coverage and evidence\".\n\nTop card:\n\"GPT-6 Astra\"\n\"Primary branch · Execution turn 5\"\nLarge right status \"Unrated\", with smaller label \"Competence\".\nBody: \"No assessment for this stage yet.\"\nBottom-right action: ONLY \"Execution evidence\". Do not add an assessment-feedback action to this unrated card.\n\nMiddle card:\n\"Qwen-3.8 Flash\"\n\"Economy branch · Execution turn 4\"\nLarge exact score \"0.48 / 1\", smaller label \"Competence\".\nBody: \"Assessment turns 3–4 · Raw score 0.485\".\nBottom-right actions: \"Execution evidence\" | \"Assessment feedback\".\n\nBottom card:\n\"GPT-6 Astra\"\n\"Primary branch · Execution turn 2\"\nLarge exact score \"0.78 / 1\", smaller label \"Competence\".\nBody: \"Assessment turn 2 · Raw score 0.775\".\nBottom-right actions: \"Execution evidence\" | \"Assessment feedback\".\n\nBottom information strip: \"Scores describe individual stages, not whole-task acceptance. Unrated stages remain visible.\"\nConstraints: Preserve every numerical value, the three cards and their order, the top unrated state, the raw scores 0.485 and 0.775, execution/assessment turn attribution, all existing icons and controls. Use no Chinese characters anywhere. Do not add data, scores, charts, thresholds, evidence-completeness badges, footers, watermarks or captions. Preserve the clean full-frame UI design and all unrelated artwork. Opaque background.",
+ "tool": "built-in image_gen",
+ "change": "English text localization; original scores, stage counts and unrated state retained."
+ },
+ {
+ "path": "assets/desktop-plan-competence-en.png",
+ "language": "en",
+ "sha256": "db20535fe5c7b96d11ac9aa60b98a8eb6b814b1f8018cea2a6a326ba805a135d",
+ "source_image": "assets/desktop-plan-competence-zh.png",
+ "source_sha256": "8dd492e844b3376709ce02276d84c581ad9580afdccd7a87c1c982a92aedb68c",
+ "edit_prompt": "Use case: text-localization.\nAsset type: English-language version of an existing HiRoute Desktop routing-plan performance UI illustration.\nInput image: the single referenced image is the edit target.\nPrimary request: Translate ALL Chinese UI text to English. Preserve the authentic full app frame, approximately 1.6:1 landscape ratio, sidebar and route-list columns, two equal metric cards, three detail rows, icons, logos, subdued grey/lavender palette, hierarchy and exact values. Keep a flat polished desktop UI with crisp legible English. Never crop. Allow minor text reflow or font-size adjustments to fit English labels without overlap or truncation.\n\nExact English UI text:\nTop window title and page heading: \"Smart Routing\".\nPage subtitle: \"Create reusable routing plans for your agents.\"\nTop-right button: \"+ New Route\".\nNavigation: \"Home\", \"Models\", \"Smart Routing\", \"Agent\", \"Sessions\", \"Settings\". Preserve \"HiRoute\" and \"AGENT ROUTING\".\nRoute list and main route title: \"Astra × Qwen · Smart Routing\".\nRoute list subline: \"Smart Saving · Research and implementation\".\nMain route strategy: \"Smart Saving\".\nGreen status badge: \"Enabled\".\nHeader buttons: \"Save Draft\", \"Publish Changes\".\nSection: \"Runtime Performance\".\nSubtitle: \"Execution stages for the selected version and time range.\"\nScope dropdowns: \"Active version r1\" and \"Last 7 days\".\n\nLeft metric card:\n\"Qwen-3.8 Flash\"\n\"Economy group\"\n\"Average stage competence\"\nLarge exact value: \"0.48 / 1\"\nCount: \"1 scored stage · 0 unrated\"\nAction: \"View stages →\".\n\nRight metric card:\n\"GPT-6 Astra\"\n\"Primary group\"\n\"Average stage competence\"\nLarge exact value: \"0.78 / 1\"\nCount: \"1 scored stage · 1 unrated\"\nAction: \"View stages →\".\n\nInformation strip below cards: \"Averages include scored stages only. Unrated stages are excluded.\"\nDetail section heading: \"Stage Details\".\nDetail-only dropdown: \"All scores\".\nThree rows in this exact order:\n\"GPT-6 Astra\" | \"Primary branch · Turn 5\" | \"Unrated\" | \"Execution evidence\"\n\"Qwen-3.8 Flash\" | \"Economy branch · Turn 4\" | \"0.48 / 1\" | \"View evidence\"\n\"GPT-6 Astra\" | \"Primary branch · Turn 2\" | \"0.78 / 1\" | \"View evidence\".\nBottom note: \"Stage performance in this route is not overall model capability or task success rate.\"\n\nConstraints: Do not change any score, sample count, model identity, version r1, 7-day filter, turn number or unrated state. Keep both model cards equally neutral rather than a leaderboard. Preserve original product/provider logos. No Chinese characters anywhere. Add no metrics, thresholds, charts, warning badges, completeness claims, watermarks or production-note captions. Preserve the full app frame and all unrelated artwork. Opaque light background.",
+ "tool": "built-in image_gen",
+ "change": "English text localization; original scores, stage counts and unrated state retained."
}
]
}
diff --git a/news/index.json b/news/index.json
index a640826b..63e6d24e 100644
--- a/news/index.json
+++ b/news/index.json
@@ -6,6 +6,10 @@
"zh": "HiRoute 智能路由实战:GPT-6 Astra × Qwen-3.8 Flash,同等质量下成本降低 90%",
"en": "HiRoute Smart Routing in Action: GPT-6 Astra × Qwen-3.8 Flash, Same Quality at 90% Lower Cost"
},
+ "shortTitle": {
+ "zh": "Astra × Qwen:同等质量,成本降低 90%",
+ "en": "Astra × Qwen: same quality, 90% lower cost"
+ },
"description": {
"zh": "用真实 Desktop 配置,看看 HiRoute 如何让 Qwen 做常规工作、Astra 接住关键决策:同等交付标准下降本超 90%,长任务还能依据执行表现自主切换模型。",
"en": "See the real Desktop setup: Qwen handles routine work, Astra handles critical decisions. Over 90% lower cost at the same delivery standard, plus an unattended coding handoff."