@goodandready/dsh-model-sync 0.5.16 → 0.5.17

package/README.md

@@ -78,6 +78,9 @@ graph LR
 
 ## ✨ Key Features
 
+* 🏷️ **Smart Collections & Catalog Auto-Tagging**: Automatically categorizes models into standardized tags (`coding`, `reasoning`, `fast`, `vision`, `long-context`) during discovery, provides fast pill-button filtering in the Web UI, and introduces a one-click "⭐ Curated Best" selection preset for top flagship models.
+* ⚡ **Latency TTFT Radar & Sparklines**: Probes response latency and Time To First Token (TTFT) across models, renders ASCII sparklines (` ▂▃▂ `) and 4-tier speed badges (⚡ `< 300ms`, 🟢 `< 1.2s`, 🟡 `1.5 - 3.0s`, 🔴 `> 3.0s`), and exposes `GET /dsh-model-sync/health/latency` for intelligent downstream routing.
+* 💸 **Dynamic Live Pricing Sync & Prompt Caching**: Periodic background updates of live model pricing from OpenRouter and LiteLLM snapshots with persistent disk fallback cache, tracking prompt cache read/write costs (`cacheReadPerToken`, `cacheWritePerToken`), with `GET /dsh-model-sync/pricing` and `POST /dsh-model-sync/pricing/sync` endpoints.
 * 💰 **Built-in Reference Pricing Database**: Canonical token pricing database automatically enriches models for OpenAI, Anthropic, Groq, Mistral, xAI, and DeepSeek, enabling cost-limit filtering and `@cheap` auto-resolution even when upstream APIs omit pricing metadata.
 * 🎯 **Dynamic Smart Virtual Aliases & Auto-Fallback**: Route requests to virtual aliases like `@fast`, `@cheap`, and `@best-code`, with automatic fallback chains when circuit breakers open or health probes fail.
 * 📏 **Real Context Limits & Boundary Safety (`contextWindow` / `maxOutputTokens`)**: precise separation of input and generation limits, prevention of 400 `max_tokens` errors via auto-clamping, and clear `In: 128k | Out: 8k` badges in Web UI.

package/README.ru.md

@@ -78,6 +78,9 @@ graph LR
 
 ## ✨ Ключевые возможности
 
+* 🏷️ **Смарт-коллекции и авто-тегирование каталога**: Автоматическое присвоение тегов моделям (`coding`, `reasoning`, `fast`, `vision`, `long-context`) при синхронизации, фильтрация по тегам pill-кнопками в Web UI и пресет «⭐ Curated Best» для выбора лучших флагманских моделей в один клик.
+* ⚡ **Радар задержек (Latency & TTFT) и спарклайны**: Измерение времени ответа и Time To First Token (TTFT), визуализация спарклайнами (` ▂▃▂ `) с 4 градациями скорости (⚡ `< 300мс`, 🟢 `< 1.2с`, 🟡 `1.5 - 3.0с`, 🔴 `> 3.0с`) и эндпоинт `GET /dsh-model-sync/health/latency` для агентов-маршрутизаторов.
+* 💸 **Динамическая синхронизация цен и учёт Prompt Caching**: Фоновое обновление актуальных тарифов из OpenRouter и LiteLLM с дисковым fallback-кэшем, поддержка скидок кэширования промптов (`cacheReadPerToken`, `cacheWritePerToken`) и эндпоинты `GET /dsh-model-sync/pricing`, `POST /dsh-model-sync/pricing/sync`.
 * 💰 **Встроенная база эталонных цен токенов**: Автоматическое обогащение моделей тарифами для OpenAI, Anthropic, Groq, Mistral, xAI и DeepSeek, обеспечивающее работу лимитов цен и псевдонима `@cheap` даже при отсутствии метаданных в API провайдера.
 * 🎯 **Динамические «умные» псевдонимы и авто-фоллбэк**: Поддержка виртуальных алиасов (`@fast`, `@cheap`, `@best-code`) и цепочек резервирования с автопереключением при размыкании Circuit Breaker.
 * 📏 **Реальные лимиты контекста и авто-клампинг (`contextWindow` / `maxOutputTokens`)**: точное разделение входного и выходного лимита токенов, защита от ошибок 400 `max_tokens` и индикация `Вход: 128k | Выход: 8k` в Web UI.

package/README.zh.md

@@ -78,6 +78,9 @@ graph LR
 
 ## ✨ 核心特性
 
+* 🏷️ **智能集合与模型自动标签**：同步时自动识别并标记模型分类标签（`coding` 编程、`reasoning` 思考推理、`fast` 极速轻量、`vision` 多模态、`long-context` 128k+ 长上下文），支持 UI 药丸胶囊快速筛选与一键精选顶级旗舰预设（"⭐ 精选最佳"）。
+* ⚡ **延迟雷达 (TTFT) 与火花线可视化**：实时测量模型响应时间与首字延迟 (TTFT)，提供火花线走势（` ▂▃▂ `）及 4 档速度徽章（⚡ `< 300ms`、🟢 `< 1.2s`、🟡 `1.5 - 3.0s`、🔴 `> 3.0s`），并开放 `GET /dsh-model-sync/health/latency` 路由接口。
+* 💸 **实时价格动态后台同步与提示词缓存**：定期从 OpenRouter 和 LiteLLM 获取最新模型单价，支持本地磁盘降级缓存，支持提示词缓存读写定价（`cacheReadPerToken`、`cacheWritePerToken`），并提供 `GET /dsh-model-sync/pricing` 和 `POST /dsh-model-sync/pricing/sync` 端点。
 * 💰 **内置基准价格数据库**：自动为 OpenAI、Anthropic、Groq、Mistral、xAI 和 DeepSeek 模型补充 Token 价格数据，即使服务商接口未返回价格，也可无缝支持成本上限过滤与 `@cheap` 别名解析。
 * 🎯 **动态智能虚拟别名与故障转移**：支持 `@fast`、`@cheap`、`@best-code` 等虚拟别名及故障转移链，在熔断器打开或健康检查失败时自动切换至备用模型。
 * 📏 **真实上下文限制与边界安全 (`contextWindow` / `maxOutputTokens`)**：精确区分输入与输出令牌限制，通过自动限制防止 400 `max_tokens` 错误，并在 Web UI 中显示 `输入: 128k | 输出: 8k` 徽章。

package/lib/client.js

@@ -65,6 +65,14 @@ window.__ModuleLoader__.load({
       '.dms-alert-ok{background:var(--dsw-alias-state-success-secondary,var(--dsw-alias-bg-layer-2));color:var(--dsw-alias-state-success-primary)}' +
       '.dms-alert-bad{background:var(--dsw-alias-state-error-secondary,var(--dsw-alias-bg-layer-2));color:var(--dsw-alias-state-error-primary)}' +
       '.dms-probe-result{padding:2px 6px;border-radius:4px;font-family:monospace;font-size:11px;border:1px solid var(--dsw-alias-border-l2);display:inline-flex;align-items:center;gap:4px}' +
+      '.dms-pill{display:inline-flex;align-items:center;gap:4px;font-size:11px;padding:3px 9px;border-radius:999px;border:1px solid var(--dsw-alias-border-l2);background:transparent;color:var(--dsw-alias-label-secondary);cursor:pointer}' +
+      '.dms-pill:hover{border-color:var(--dsw-alias-brand-primary);color:var(--dsw-alias-label-primary)}' +
+      '.dms-pill-active{background:var(--dsw-alias-label-primary)!important;color:var(--dsw-alias-bg-layer-3)!important;border-color:transparent!important;font-weight:600}' +
+      '.dms-tier-lightning{border-color:#a855f7;color:#a855f7;background:rgba(168,85,247,0.1)}' +
+      '.dms-tier-fast{border-color:var(--dsw-alias-state-success-primary);color:var(--dsw-alias-state-success-primary);background:var(--dsw-alias-state-success-secondary,rgba(34,197,94,0.1))}' +
+      '.dms-tier-moderate{border-color:var(--dsw-alias-state-warn-primary);color:var(--dsw-alias-state-warn-primary);background:var(--dsw-alias-state-warn-secondary,rgba(234,179,8,0.1))}' +
+      '.dms-tier-slow{border-color:var(--dsw-alias-state-error-primary);color:var(--dsw-alias-state-error-primary);background:var(--dsw-alias-state-error-secondary,rgba(239,68,68,0.1))}' +
+      '.dms-sparkline{font-family:monospace;letter-spacing:-1px;margin-left:3px;font-weight:bold}' +
       '@media (max-width:540px){.dms-provider{grid-template-columns:1fr!important}.dms-provider .dms-btn{justify-self:start}.dms-toolbar>div:first-child{align-items:stretch;flex-direction:column}.dms-toolbar>div:first-child select{width:100%}.dms-toolbar,.dms-section,.dms-section-card,.dms-editor{padding:11px 12px}}'
 
     function createErrorBoundary() {
@@ -228,6 +236,18 @@ window.__ModuleLoader__.load({
       notifyOnPriceChanged: 'Notify on price changes',
       saveWebhook: 'Save Webhook Settings',
       testWebhook: 'Test Webhook',
+      selectCurated: '⭐ Curated Best',
+      tagAll: 'All',
+      tagCurated: '⭐ Curated',
+      tagReasoning: '🧠 Reasoning',
+      tagCoding: '💻 Coding',
+      tagFast: '⚡ Fast',
+      tagVision: '👁️ Vision',
+      tagLongContext: '📜 128k+',
+      speedRadar: 'Latency Radar',
+      ttft: 'TTFT',
+      pricingSync: 'Sync Prices',
+      pricingSyncing: 'Syncing prices…',
       webhookTesting: 'Testing...',
       webhookSuccess: 'Webhook test succeeded',
       webhookFailed: 'Webhook test failed: %s',
@@ -400,6 +420,18 @@ window.__ModuleLoader__.load({
       notifyOnPriceChanged: '模型价格变更时通知',
       saveWebhook: '保存 Webhook 设置',
       testWebhook: '测试 Webhook',
+      selectCurated: '⭐ 精选最佳',
+      tagAll: '全部',
+      tagCurated: '⭐ 精选',
+      tagReasoning: '🧠 深度思考',
+      tagCoding: '💻 编程优化',
+      tagFast: '⚡ 极速轻量',
+      tagVision: '👁️ 多模态视觉',
+      tagLongContext: '📜 长上下文',
+      speedRadar: '延迟雷达',
+      ttft: '首字延迟',
+      pricingSync: '同步价格',
+      pricingSyncing: '正在同步价格…',
       webhookTesting: '正在测试...',
       webhookSuccess: 'Webhook 测试成功',
       webhookFailed: 'Webhook 测试失败：%s',
@@ -473,6 +505,81 @@ window.__ModuleLoader__.load({
     }
 
     
+    const SPARKLINE_BARS = [' ', '▂', '▃', '▄', '▅', '▆', '▇', '█']
+    function generateSparkline(values) {
+      if (!Array.isArray(values) || values.length === 0) return ''
+      const nums = values.filter((v) => typeof v === 'number' && Number.isFinite(v))
+      if (nums.length === 0) return ''
+      if (nums.length === 1) return '▄'
+      const min = Math.min(...nums)
+      const max = Math.max(...nums)
+      if (min === max) return '▄'.repeat(nums.length)
+      const range = max - min
+      return nums.map((v) => SPARKLINE_BARS[Math.min(7, Math.max(0, Math.floor(((v - min) / range) * 7)))]).join('')
+    }
+
+    function classifySpeedTier(latencyMs, isOk = true) {
+      if (!isOk) return 'degraded'
+      if (typeof latencyMs !== 'number' || !Number.isFinite(latencyMs) || latencyMs <= 0) return 'unknown'
+      if (latencyMs < 300) return 'lightning'
+      if (latencyMs < 1200) return 'fast'
+      if (latencyMs <= 3000) return 'moderate'
+      return 'slow'
+    }
+
+    function deriveModelTags(model) {
+      if (!model || typeof model !== 'object') return []
+      const tags = new Set(Array.isArray(model.tags) ? model.tags.filter((t) => typeof t === 'string' && t.trim()).map((t) => t.trim().toLowerCase()) : [])
+      const idLower = String(model.id || '').toLowerCase()
+      const nameLower = String(model.name || '').toLowerCase()
+      const provLower = String(model.provider || '').toLowerCase()
+      const capabilities = model.capabilities || {}
+
+      if (capabilities.code === true || idLower.includes('coder') || idLower.includes('codestral') || idLower.includes('starcoder') || idLower.includes('deepseek-coder') || idLower.includes('claude-3-7-sonnet') || idLower.includes('claude-3-5-sonnet') || idLower.includes('kimi-k2.7-code') || idLower.includes('qwen-2.5-coder') || idLower.includes('qwen2.5-coder') || (/(^|[_-])(code|dev|prog)($|[_-])/i.test(idLower)) || (/(^|[_-])(code|coder)($|[_-])/i.test(nameLower))) {
+        tags.add('coding')
+      }
+      if (model.reasoningConfig?.supportsReasoning === true || capabilities.reasoning === true || /(^|[_-])(o1|o3|r1|qwq)($|[_-])/i.test(idLower) || idLower.includes('reasoner') || idLower.includes('deepseek-r1') || idLower.includes('thinking') || idLower.includes('-r1')) {
+        tags.add('reasoning')
+      }
+      if (provLower === 'groq' || provLower === 'cerebras' || idLower.includes('flash') || idLower.includes('mini') || idLower.includes('instant') || idLower.includes('haiku') || idLower.includes('small') || idLower.includes('8b') || idLower.includes('3b') || idLower.includes('1b') || idLower.includes('turbo') || idLower.includes('lite') || idLower.includes('nano')) {
+        tags.add('fast')
+      }
+      if (capabilities.vision === true || (Array.isArray(model.input) && model.input.includes('image')) || idLower.includes('vision') || idLower.includes('vl') || idLower.includes('pixtral') || idLower.includes('gemini') || idLower.includes('claude-3') || idLower.includes('4o')) {
+        tags.add('vision')
+      }
+      if (typeof model.contextWindow === 'number' && model.contextWindow >= 128000) {
+        tags.add('long-context')
+      }
+      return [...tags].sort()
+    }
+
+    function isCuratedModel(model) {
+      if (!model || typeof model !== 'object') return false
+      if (model.lifecycle?.deprecated) return false
+      const id = String(model.id || '').toLowerCase()
+      if (/-(202\d{5}|202\d-\d{2}-\d{2}|\d{4})$/.test(id) && !id.includes('claude-3-7') && !id.includes('claude-3-5')) {
+        return false
+      }
+      if (id.includes('-0314') || id.includes('-0613') || id.includes('-1106') || id.includes('-0125')) {
+        return false
+      }
+      const curatedIds = [
+        'claude-3-7-sonnet', 'claude-3-5-sonnet', 'claude-3-5-haiku',
+        'gpt-4o', 'gpt-4o-mini', 'o1', 'o3-mini',
+        'deepseek-chat', 'deepseek-reasoner', 'deepseek-v3', 'deepseek-r1',
+        'gemini-2.0-flash', 'gemini-1.5-pro', 'gemini-1.5-flash',
+        'llama-3.3-70b-versatile', 'llama-3.1-8b-instant', 'llama-3.1-70b-versatile',
+        'qwen-2.5-coder-32b', 'qwen-2.5-72b', 'qwq-32b',
+        'mistral-large', 'codestral', 'pixtral-large',
+        'kimi-k2.7-code'
+      ]
+      for (const c of curatedIds) {
+        if (id === c || id.endsWith('/' + c) || id.includes(c)) return true
+      }
+      const tags = model.tags || deriveModelTags(model)
+      return tags.includes('coding') && tags.includes('reasoning') && tags.includes('long-context')
+    }
+
     function formatTokens(num) {
       if (typeof num !== "number" || !Number.isFinite(num) || num <= 0) return "?";
       if (num >= 1000000) {
@@ -680,6 +787,7 @@ window.__ModuleLoader__.load({
       })
       const [pageSize, setPageSize] = React.useState(50)
       const [quickFilter, setQuickFilter] = React.useState({ context100k: false, cheap: false })
+      const [selectedTag, setSelectedTag] = React.useState('all')
       const [tryState, setTryState] = React.useState({})
       const [batchTrying, setBatchTrying] = React.useState(false)
       const [batchSummary, setBatchSummary] = React.useState(null)
@@ -1273,7 +1381,15 @@ window.__ModuleLoader__.load({
           const haystack = (m.id + ' ' + (m.name || '') + ' ' + (m.aliases || []).join(' ') + ' ' + (m.tags || []).join(' ') + ' ' + (m.description || '')).toLowerCase()
           return tokens.every((tok) => haystack.includes(tok))
         }) : filteredChoicesBase
-        const quickFilteredChoices = searchedChoices.filter((m) => {
+        const taggedChoices = selectedTag === 'all'
+          ? searchedChoices
+          : selectedTag === 'curated'
+            ? searchedChoices.filter((m) => isCuratedModel(m))
+            : searchedChoices.filter((m) => {
+                const tags = (m.tags && m.tags.length > 0) ? m.tags : deriveModelTags(m)
+                return tags.includes(selectedTag)
+              })
+        const quickFilteredChoices = taggedChoices.filter((m) => {
           if (quickFilter.context100k && (!m.contextWindow || m.contextWindow < 100000)) return false
           if (quickFilter.cheap) {
             const price = m.pricing?.inputPerToken
@@ -1296,7 +1412,7 @@ window.__ModuleLoader__.load({
           }
           return (a.name || a.id || '').localeCompare(b.name || b.id || '')
         })
-      }, [choices, capabilityFilter, debouncedSearchQuery, quickFilter, sortBy])
+      }, [choices, capabilityFilter, debouncedSearchQuery, selectedTag, quickFilter, sortBy])
 
       const maxContext = React.useMemo(() => filteredChoices.reduce((max, m) => Math.max(max, m.contextWindow || 0), 0), [filteredChoices])
       const minPrice = React.useMemo(() => filteredChoices.reduce((min, m) => m.pricing?.inputPerToken !== undefined ? Math.min(min, m.pricing.inputPerToken) : min, Infinity), [filteredChoices])
@@ -1700,6 +1816,13 @@ window.__ModuleLoader__.load({
                   ))
                 ) : React.createElement('small', { style: { display: 'block', marginBottom: 8 } }, t.noCapabilities),
                 React.createElement('div', { style: { display: 'flex', gap: 6, margin: '6px 0', flexWrap: 'wrap' } },
+                  button(t.selectCurated || '⭐ Curated Best', () => {
+                    const ids = new Set(selectionDraft)
+                    for (const m of choices) {
+                      if (isCuratedModel(m)) ids.add(m.id)
+                    }
+                    setSelectionDraft([...ids])
+                  }, 'primary'),
                   button(t.selectAllFiltered, () => { const ids = new Set(selectionDraft); for (const m of filteredChoices) ids.add(m.id); setSelectionDraft([...ids]) }),
                   button(t.deselectAllFiltered, () => { const filterSet = new Set(filteredChoices.map((m) => m.id)); setSelectionDraft(selectionDraft.filter((id) => !filterSet.has(id))) }),
                   button(t.invertFiltered, () => { const current = new Set(selectionDraft); for (const m of filteredChoices) { if (current.has(m.id)) current.delete(m.id); else current.add(m.id); } setSelectionDraft([...current]) })
@@ -1710,6 +1833,22 @@ window.__ModuleLoader__.load({
                   batchSummary ? React.createElement('small', { style: { color: 'var(--dsw-alias-label-secondary)' } }, t.batchTryResults.replace('%d', batchSummary.reachable).replace('%d', batchSummary.total)) : null
                 ),
                 React.createElement('small', { style: { display: 'block', marginBottom: 8 } }, `${filteredChoices.length}/${choices.length} ${t.capabilityCount}`),
+                React.createElement('div', { style: { display: 'flex', gap: 6, margin: '8px 0', flexWrap: 'wrap', alignItems: 'center' } },
+                  [
+                    { id: 'all', label: t.tagAll || 'All' },
+                    { id: 'curated', label: t.tagCurated || '⭐ Curated' },
+                    { id: 'reasoning', label: t.tagReasoning || '🧠 Reasoning' },
+                    { id: 'coding', label: t.tagCoding || '💻 Coding' },
+                    { id: 'fast', label: t.tagFast || '⚡ Fast' },
+                    { id: 'vision', label: t.tagVision || '👁️ Vision' },
+                    { id: 'long-context', label: t.tagLongContext || '📜 128k+' },
+                  ].map((pill) => React.createElement('button', {
+                    key: pill.id,
+                    type: 'button',
+                    className: 'dms-pill' + (selectedTag === pill.id ? ' dms-pill-active' : ''),
+                    onClick: () => setSelectedTag(pill.id),
+                  }, pill.label))
+                ),
                 visibleChoices.map((model) => {
                   const deprecationDays = (() => {
                     const d = model.lifecycle?.deprecationDate
@@ -1717,7 +1856,8 @@ window.__ModuleLoader__.load({
                     const diff = new Date(d).getTime() - Date.now()
                     return diff > 0 ? Math.ceil(diff / (1000 * 60 * 60 * 24)) : 0
                   })()
-                  const probe = tryState[`${row.provider}:${model.id}`]
+                  const radar = status?.latencyRadar?.radar?.find((r) => r.provider === row.provider && r.model === model.id)
+                  const probe = tryState[`${row.provider}:${model.id}`] || (radar ? { ok: radar.ok, latency: radar.latencyMs, ttftMs: radar.ttftMs, sparkline: radar.sparkline, status: radar.status } : null)
                   return React.createElement('label', { key: model.id, style: { display: 'block', margin: '4px 0', padding: '4px 6px', borderRadius: 6, background: isNewModel(model, row) ? 'var(--dsw-alias-bg-layer-2)' : 'transparent' } },
                     React.createElement('input', { type: 'checkbox', checked: selectionDraft.includes(model.id), onChange: (e) => setSelectionDraft((current) => e.target.checked ? [...current, model.id] : current.filter((id) => id !== model.id)) }),
                     ` ${model.name ?? model.id} (${model.id})`,
@@ -1749,12 +1889,36 @@ window.__ModuleLoader__.load({
                     }, `🧠 ${model.reasoningConfig?.paramFormat || 'reasoning'}`) : null,
                     (model.contextWindow && maxContext > 0 && model.contextWindow >= maxContext) ? React.createElement('span', { style: { marginLeft: 4, fontSize: 10, padding: '1px 4px', borderRadius: 3, background: 'var(--dsw-alias-state-warn-primary)', color: 'white' } }, '★') : null,
                     (model.pricing && minPrice !== Infinity && model.pricing.inputPerToken === minPrice) ? React.createElement('span', { style: { marginLeft: 4, fontSize: 10, padding: '1px 4px', borderRadius: 3, background: 'var(--dsw-alias-state-success-primary)', color: 'white' } }, '$') : null,
-                    model.tags && model.tags.length ? React.createElement('small', { style: { marginLeft: 6, color: 'var(--dsw-alias-label-secondary)' } }, model.tags.join(', ')) : null,
+                    (() => {
+                      const modelTags = (model.tags && model.tags.length > 0) ? model.tags : deriveModelTags(model)
+                      if (!modelTags || modelTags.length === 0) return null
+                      return React.createElement('span', { style: { marginLeft: 6, display: 'inline-flex', gap: 3, verticalAlign: 'middle' } },
+                        modelTags.map((tag) => React.createElement('span', {
+                          key: tag,
+                          style: {
+                            fontSize: 10,
+                            padding: '0 4px',
+                            borderRadius: 3,
+                            background: 'var(--dsw-alias-bg-layer-2)',
+                            border: '1px solid var(--dsw-alias-border-l1)',
+                            color: 'var(--dsw-alias-label-secondary)',
+                          }
+                        }, tag === 'coding' ? '💻 coding' : tag === 'reasoning' ? '🧠 reasoning' : tag === 'fast' ? '⚡ fast' : tag === 'vision' ? '👁️ vision' : tag === 'long-context' ? '📜 128k+' : tag))
+                      )
+                    })(),
                     model.pricing ? React.createElement('small', { style: { marginLeft: 6, color: 'var(--dsw-alias-label-secondary)', fontSize: 11 } }, `in $${(model.pricing.inputPerToken * 1000000).toFixed(2)}/1M` + (model.pricing.cacheReadPerToken !== undefined ? ` / cache $${(model.pricing.cacheReadPerToken * 1000000).toFixed(2)}` : '') + (model.pricing.outputPerToken !== undefined ? ` / out $${(model.pricing.outputPerToken * 1000000).toFixed(2)}` : '')) : null,
                     model.description ? React.createElement('small', { title: model.description, style: { display: 'block', marginLeft: 22, color: 'var(--dsw-alias-label-secondary)', fontSize: 11 } }, model.description.slice(0, 120)) : null,
                     row.baseURL || row.api ? React.createElement('a', { href: (row.baseURL || '') + '/models/' + encodeURIComponent(model.id), target: '_blank', rel: 'noopener', style: { marginLeft: 8, fontSize: 11 } }, 'docs') : null,
                     React.createElement('button', { type: 'button', onClick: (e) => { e.preventDefault(); doTry(row.provider, model.id) }, style: { marginLeft: 8, fontSize: 11, padding: '1px 6px', borderRadius: 4 } }, (probe && probe.loading) ? '…' : '▶'),
-                    probe && !probe.loading ? React.createElement('span', { className: 'dms-badge ' + (probe.ok ? 'dms-badge-ok' : 'dms-badge-bad'), style: { marginLeft: 6 } }, probe.ok ? `${probe.latency}ms` : (probe.error || probe.status || 'offline')) : null
+                    probe && !probe.loading ? React.createElement('span', {
+                      className: 'dms-badge ' + (probe.ok ? (classifySpeedTier(probe.latency, probe.ok) === 'lightning' ? 'dms-tier-lightning' : classifySpeedTier(probe.latency, probe.ok) === 'fast' ? 'dms-tier-fast' : 'dms-tier-moderate') : 'dms-badge-bad'),
+                      style: { marginLeft: 6 },
+                      title: probe.ttftMs ? `TTFT: ${probe.ttftMs}ms | Latency: ${probe.latency}ms` : (probe.latency ? `${probe.latency}ms` : undefined)
+                    },
+                      (classifySpeedTier(probe.latency, probe.ok) === 'lightning' ? '⚡ ' : classifySpeedTier(probe.latency, probe.ok) === 'fast' ? '🟢 ' : classifySpeedTier(probe.latency, probe.ok) === 'moderate' ? '🟡 ' : '🔴 ') +
+                      (probe.ok ? `${probe.latency}ms` : (probe.error || probe.status || 'offline')) +
+                      (probe.sparkline ? ` ${probe.sparkline}` : '')
+                    ) : null
                   )
                 }),
                 filteredChoices.length > pageSize ? React.createElement('div', { style: { display: 'flex', alignItems: 'center', gap: 8, margin: '8px 0' } },
@@ -1916,7 +2080,7 @@ window.__ModuleLoader__.load({
       trySlot('settings.plugin.item', () => ctx.slots.register({ name: 'settings.plugin.item', key: NS, locale: NS, inject: () => ({ ctx }) }, (props) => React.createElement(ErrorBoundary, null, React.createElement(ModelSyncCard, { ...props, ctx }))))
     }
 
-    module.exports = { apply, inject: ['slots', 'locale', 'configForms'], filterModelsByCapabilities }
+    module.exports = { apply, inject: ['slots', 'locale', 'configForms'], filterModelsByCapabilities, deriveModelTags, isCuratedModel, generateSparkline, classifySpeedTier }
     return module.exports
   },
 })

package/lib/http.js

@@ -339,7 +339,46 @@ export function registerHttpApi(ctx, synchronizer, { getSchedulerStatus = () =>
       }
     }
     if (route === 'import') {
-      if (req.method !== 'POST') return json(res, 405, { error: { code: 'method', message: 'POST only' } })
+      if (route === 'health-latency') {
+      if (req.method !== 'GET') return json(res, 405, { error: { code: 'method', message: 'GET only' } })
+      try {
+        const urlObj = new URL(req.url, 'http://127.0.0.1')
+        const provider = urlObj.searchParams.get('provider') || undefined
+        const report = typeof synchronizer.getLatencyRadar === 'function'
+          ? synchronizer.getLatencyRadar({ provider })
+          : { updatedAt: Date.now(), radar: [] }
+        return json(res, 200, report)
+      } catch (error) {
+        return json(res, 400, { error: { code: 'bad-request', message: error instanceof Error ? error.message : String(error) } })
+      }
+    }
+    if (route === 'pricing') {
+      if (req.method === 'GET') {
+        try {
+          const urlObj = new URL(req.url, 'http://127.0.0.1')
+          const provider = urlObj.searchParams.get('provider') || undefined
+          const model = urlObj.searchParams.get('model') || undefined
+          const report = typeof synchronizer.getPricingInfo === 'function'
+            ? synchronizer.getPricingInfo({ provider, model })
+            : { pricing: {} }
+          return json(res, 200, report)
+        } catch (error) {
+          return json(res, 400, { error: { code: 'bad-request', message: error instanceof Error ? error.message : String(error) } })
+        }
+      }
+      if (req.method === 'POST') {
+        try {
+          const result = typeof synchronizer.syncPricing === 'function'
+            ? await synchronizer.syncPricing()
+            : { updated: 0, source: 'reference' }
+          return json(res, 200, result)
+        } catch (error) {
+          return json(res, 500, { error: { code: 'sync-failed', message: error instanceof Error ? error.message : String(error) } })
+        }
+      }
+      return json(res, 405, { error: { code: 'method', message: 'GET or POST only' } })
+    }
+    if (req.method !== 'POST') return json(res, 405, { error: { code: 'method', message: 'POST only' } })
       try {
         const body = await readJson(req)
         const result = await synchronizer.importConfig(body)
@@ -357,6 +396,45 @@ export function registerHttpApi(ctx, synchronizer, { getSchedulerStatus = () =>
       if (req.method !== 'GET') return json(res, 405, { error: { code: 'method', message: 'GET only' } })
       return json(res, 200, { notifications: synchronizer.notifications(normalizeNotificationRequest(req.url)) })
     }
+    if (route === 'health-latency') {
+      if (req.method !== 'GET') return json(res, 405, { error: { code: 'method', message: 'GET only' } })
+      try {
+        const urlObj = new URL(req.url, 'http://127.0.0.1')
+        const provider = urlObj.searchParams.get('provider') || undefined
+        const report = typeof synchronizer.getLatencyRadar === 'function'
+          ? synchronizer.getLatencyRadar({ provider })
+          : { updatedAt: Date.now(), radar: [] }
+        return json(res, 200, report)
+      } catch (error) {
+        return json(res, 400, { error: { code: 'bad-request', message: error instanceof Error ? error.message : String(error) } })
+      }
+    }
+    if (route === 'pricing') {
+      if (req.method === 'GET') {
+        try {
+          const urlObj = new URL(req.url, 'http://127.0.0.1')
+          const provider = urlObj.searchParams.get('provider') || undefined
+          const model = urlObj.searchParams.get('model') || undefined
+          const report = typeof synchronizer.getPricingInfo === 'function'
+            ? synchronizer.getPricingInfo({ provider, model })
+            : { pricing: {} }
+          return json(res, 200, report)
+        } catch (error) {
+          return json(res, 400, { error: { code: 'bad-request', message: error instanceof Error ? error.message : String(error) } })
+        }
+      }
+      if (req.method === 'POST') {
+        try {
+          const result = typeof synchronizer.syncPricing === 'function'
+            ? await synchronizer.syncPricing()
+            : { updated: 0, source: 'reference' }
+          return json(res, 200, result)
+        } catch (error) {
+          return json(res, 500, { error: { code: 'sync-failed', message: error instanceof Error ? error.message : String(error) } })
+        }
+      }
+      return json(res, 405, { error: { code: 'method', message: 'GET or POST only' } })
+    }
     if (req.method !== 'POST') return json(res, 405, { error: { code: 'method', message: 'POST only' } })
     let options
     try {
@@ -423,5 +501,7 @@ export function registerHttpApi(ctx, synchronizer, { getSchedulerStatus = () =>
     ctx.webServer.register({ kind: 'exact', path: BASE + '/notifications/read', handler: makeHandler('notification-read') }),
     ctx.webServer.register({ kind: 'exact', path: BASE + '/notifications/acknowledge', handler: makeHandler('notification-acknowledge') }),
     ctx.webServer.register({ kind: 'exact', path: BASE + '/webhook/test', handler: makeHandler('webhook-test') }),
+    ctx.webServer.register({ kind: 'exact', path: BASE + '/health/latency', handler: makeHandler('health-latency') }),
+    ctx.webServer.register({ kind: 'exact', path: BASE + '/pricing', handler: makeHandler('pricing') }),
   ]
 }

package/lib/models.js

@@ -365,6 +365,120 @@ function normalizeReasoningConfig(provider, row, id, capabilities) {
   return source
 }
 
+export function deriveModelTags(model) {
+  if (!model || typeof model !== "object") return []
+  const tags = new Set(Array.isArray(model.tags) ? model.tags.filter((t) => typeof t === "string" && t.trim()).map((t) => t.trim().toLowerCase()) : [])
+  const idLower = String(model.id || "").toLowerCase()
+  const nameLower = String(model.name || "").toLowerCase()
+  const provLower = String(model.provider || "").toLowerCase()
+  const capabilities = model.capabilities || {}
+
+  // 1. coding
+  if (
+    capabilities.code === true ||
+    idLower.includes("coder") ||
+    idLower.includes("codestral") ||
+    idLower.includes("starcoder") ||
+    idLower.includes("deepseek-coder") ||
+    idLower.includes("claude-3-7-sonnet") ||
+    idLower.includes("claude-3-5-sonnet") ||
+    idLower.includes("kimi-k2.7-code") ||
+    idLower.includes("qwen-2.5-coder") ||
+    idLower.includes("qwen2.5-coder") ||
+    (/(^|[_-])(code|dev|prog)($|[_-])/i.test(idLower)) ||
+    (/(^|[_-])(code|coder)($|[_-])/i.test(nameLower))
+  ) {
+    tags.add("coding")
+  }
+
+  // 2. reasoning
+  if (
+    model.reasoningConfig?.supportsReasoning === true ||
+    capabilities.reasoning === true ||
+    /(^|[_-])(o1|o3|r1|qwq)($|[_-])/i.test(idLower) ||
+    idLower.includes("reasoner") ||
+    idLower.includes("deepseek-r1") ||
+    idLower.includes("thinking") ||
+    idLower.includes("-r1")
+  ) {
+    tags.add("reasoning")
+  }
+
+  // 3. fast
+  if (
+    provLower === "groq" ||
+    provLower === "cerebras" ||
+    idLower.includes("flash") ||
+    idLower.includes("mini") ||
+    idLower.includes("instant") ||
+    idLower.includes("haiku") ||
+    idLower.includes("small") ||
+    idLower.includes("8b") ||
+    idLower.includes("3b") ||
+    idLower.includes("1b") ||
+    idLower.includes("turbo") ||
+    idLower.includes("lite") ||
+    idLower.includes("nano")
+  ) {
+    tags.add("fast")
+  }
+
+  // 4. vision
+  if (
+    capabilities.vision === true ||
+    (Array.isArray(model.input) && model.input.includes("image")) ||
+    idLower.includes("vision") ||
+    idLower.includes("vl") ||
+    idLower.includes("pixtral") ||
+    idLower.includes("gemini") ||
+    idLower.includes("claude-3") ||
+    idLower.includes("4o")
+  ) {
+    tags.add("vision")
+  }
+
+  // 5. long-context (>= 128k tokens)
+  if (typeof model.contextWindow === "number" && model.contextWindow >= 128000) {
+    tags.add("long-context")
+  }
+
+  return [...tags].sort()
+}
+
+export function isCuratedModel(model) {
+  if (!model || typeof model !== "object") return false
+  if (model.lifecycle?.deprecated) return false
+  const id = String(model.id || "").toLowerCase()
+  if (/-(202\d{5}|202\d-\d{2}-\d{2}|\d{4})$/.test(id) && !id.includes("claude-3-7") && !id.includes("claude-3-5")) {
+    return false
+  }
+  if (id.includes("-0314") || id.includes("-0613") || id.includes("-1106") || id.includes("-0125")) {
+    return false
+  }
+
+  const curatedIds = [
+    "claude-3-7-sonnet", "claude-3-5-sonnet", "claude-3-5-haiku",
+    "gpt-4o", "gpt-4o-mini", "o1", "o3-mini",
+    "deepseek-chat", "deepseek-reasoner", "deepseek-v3", "deepseek-r1",
+    "gemini-2.0-flash", "gemini-1.5-pro", "gemini-1.5-flash",
+    "llama-3.3-70b-versatile", "llama-3.1-8b-instant", "llama-3.1-70b-versatile",
+    "qwen-2.5-coder-32b", "qwen-2.5-72b", "qwq-32b",
+    "mistral-large", "codestral", "pixtral-large",
+    "kimi-k2.7-code"
+  ]
+
+  for (const c of curatedIds) {
+    if (id === c || id.endsWith("/" + c) || id.includes(c)) return true
+  }
+
+  const tags = model.tags || deriveModelTags(model)
+  if (tags.includes("coding") && tags.includes("reasoning") && tags.includes("long-context")) {
+    return true
+  }
+
+  return false
+}
+
 export function normalizeModel(provider, row) {
   if (typeof row === 'string') {
     row = { id: row }

package/lib/pricing-database.js

@@ -1,6 +1,13 @@
-function cleanTokenPrice(val) {
-  if (typeof val !== 'number' || !Number.isFinite(val)) return 0
-  return Math.round((val / 1000000) * 1e12) / 1e12
+import fs from 'node:fs'
+import path from 'node:path'
+
+function cleanTokenPrice(val, { isPerToken = false } = {}) {
+  const num = typeof val === 'string' ? Number(val.trim()) : val
+  if (typeof num !== 'number' || !Number.isFinite(num) || num < 0) return 0
+  if (isPerToken) {
+    return Math.round(num * 1e12) / 1e12
+  }
+  return Math.round((num / 1000000) * 1e12) / 1e12
 }
 
 function perMillion(inputUSD, outputUSD, extra = {}) {
@@ -21,23 +28,23 @@ function perMillion(inputUSD, outputUSD, extra = {}) {
 
 export const REFERENCE_PRICING = Object.freeze({
   openai: {
-    'gpt-4o': perMillion(2.5, 10),
-    'gpt-4o-mini': perMillion(0.15, 0.6),
+    'gpt-4o': perMillion(2.5, 10, { cacheReadUSD: 1.25 }),
+    'gpt-4o-mini': perMillion(0.15, 0.6, { cacheReadUSD: 0.075 }),
     'gpt-4-turbo': perMillion(10, 30),
     'gpt-4': perMillion(30, 60),
     'gpt-3.5-turbo': perMillion(0.5, 1.5),
-    'o1': perMillion(15, 60),
+    'o1': perMillion(15, 60, { cacheReadUSD: 7.5 }),
     'o1-preview': perMillion(15, 60),
-    'o1-mini': perMillion(1.1, 4.4),
-    'o3-mini': perMillion(1.1, 4.4),
+    'o1-mini': perMillion(1.1, 4.4, { cacheReadUSD: 0.55 }),
+    'o3-mini': perMillion(1.1, 4.4, { cacheReadUSD: 0.55 }),
     'chatgpt-4o-latest': perMillion(5, 15),
   },
   anthropic: {
-    'claude-3-7-sonnet': perMillion(3, 15),
-    'claude-3-5-sonnet': perMillion(3, 15),
-    'claude-3-5-haiku': perMillion(0.8, 4),
-    'claude-3-opus': perMillion(15, 75),
-    'claude-3-haiku': perMillion(0.25, 1.25),
+    'claude-3-7-sonnet': perMillion(3, 15, { cacheReadUSD: 0.3, cacheWriteUSD: 3.75 }),
+    'claude-3-5-sonnet': perMillion(3, 15, { cacheReadUSD: 0.3, cacheWriteUSD: 3.75 }),
+    'claude-3-5-haiku': perMillion(0.8, 4, { cacheReadUSD: 0.08, cacheWriteUSD: 1 }),
+    'claude-3-opus': perMillion(15, 75, { cacheReadUSD: 1.5, cacheWriteUSD: 18.75 }),
+    'claude-3-haiku': perMillion(0.25, 1.25, { cacheReadUSD: 0.025, cacheWriteUSD: 0.3 }),
   },
   groq: {
     'llama-3.3-70b-versatile': perMillion(0.59, 0.79),
@@ -83,6 +90,22 @@ export const REFERENCE_PRICING = Object.freeze({
   },
 })
 
+// Dynamic in-memory pricing store populated from OpenRouter / LiteLLM / disk cache
+const DYNAMIC_PRICING = new Map()
+
+export function getDynamicPricingMap() {
+  return DYNAMIC_PRICING
+}
+
+export function clearDynamicPricing() {
+  DYNAMIC_PRICING.clear()
+}
+
+export function setDynamicPricing(key, pricing) {
+  if (!key || typeof key !== 'string' || !pricing) return
+  DYNAMIC_PRICING.set(key.toLowerCase().trim(), Object.freeze({ ...pricing }))
+}
+
 function cleanId(id) {
   if (typeof id !== 'string') return ''
   return id
@@ -100,16 +123,23 @@ export function lookupReferencePricing(provider, modelId) {
   const rawId = modelId.toLowerCase().trim()
   const cleaned = cleanId(rawId)
 
-  // 1. Direct check in provider table (exact match or date-cleaned match)
+  // 1. Dynamic pricing cache check (provider:modelId or rawId or cleaned)
+  if (DYNAMIC_PRICING.size > 0) {
+    const dynScoped = provKey ? DYNAMIC_PRICING.get(`${provKey}:${rawId}`) || DYNAMIC_PRICING.get(`${provKey}:${cleaned}`) : undefined
+    if (dynScoped) return { ...dynScoped }
+    const dynRaw = DYNAMIC_PRICING.get(rawId) || DYNAMIC_PRICING.get(cleaned)
+    if (dynRaw) return { ...dynRaw }
+  }
+
+  // 2. Direct check in provider table (exact match or date-cleaned match)
   const table = REFERENCE_PRICING[provKey]
   if (table) {
     if (table[rawId]) return { ...table[rawId] }
     if (table[cleaned]) return { ...table[cleaned] }
-    // If the provider has a dedicated pricing table, do not search other providers' tables
     return undefined
   }
 
-  // 2. Vendor prefix check (e.g. "openai/gpt-4o" or "anthropic/claude-3-5-sonnet")
+  // 3. Vendor prefix check (e.g. "openai/gpt-4o" or "anthropic/claude-3-5-sonnet")
   const slashIdx = rawId.indexOf('/')
   if (slashIdx > 0) {
     const subProv = rawId.slice(0, slashIdx)
@@ -122,10 +152,113 @@ export function lookupReferencePricing(provider, modelId) {
     }
   }
 
-  // 3. Fallback search across tables ONLY for unknown/aggregator providers without a dedicated table
+  // 4. Fallback search across tables ONLY for unknown/aggregator providers without a dedicated table
   for (const [p, t] of Object.entries(REFERENCE_PRICING)) {
     if (t[cleaned]) return { ...t[cleaned] }
   }
 
   return undefined
 }
+
+export function parseOpenRouterPricing(modelsData) {
+  if (!Array.isArray(modelsData)) return []
+  const results = []
+  for (const item of modelsData) {
+    if (!item?.id || !item?.pricing) continue
+    const id = String(item.id).trim().toLowerCase()
+    const p = item.pricing
+    const inputPerToken = cleanTokenPrice(p.prompt, { isPerToken: true })
+    const outputPerToken = cleanTokenPrice(p.completion, { isPerToken: true })
+    const cacheReadPerToken = p.input_cache_read !== undefined || p.prompt_cached !== undefined
+      ? cleanTokenPrice(p.input_cache_read ?? p.prompt_cached, { isPerToken: true })
+      : undefined
+    const cacheWritePerToken = p.input_cache_write !== undefined
+      ? cleanTokenPrice(p.input_cache_write, { isPerToken: true })
+      : undefined
+
+    if (inputPerToken > 0 || outputPerToken > 0) {
+      const entry = {
+        inputPerToken,
+        outputPerToken,
+        ...(cacheReadPerToken !== undefined ? { cacheReadPerToken } : {}),
+        ...(cacheWritePerToken !== undefined ? { cacheWritePerToken } : {}),
+        currency: 'USD',
+        source: 'openrouter',
+      }
+      results.push({ id, entry })
+    }
+  }
+  return results
+}
+
+export async function fetchPricingUpdates({
+  timeoutMs = 15000,
+  fetchFn = globalThis.fetch,
+  cachePath,
+  logger,
+} = {}) {
+  let updatedCount = 0
+  const controller = new AbortController()
+  const timer = setTimeout(() => controller.abort(), timeoutMs)
+
+  try {
+    const res = await fetchFn('https://openrouter.ai/api/v1/models', {
+      signal: controller.signal,
+      headers: { 'Accept': 'application/json' },
+    })
+    if (res.ok) {
+      const json = await res.json()
+      const entries = parseOpenRouterPricing(json?.data)
+      for (const { id, entry } of entries) {
+        DYNAMIC_PRICING.set(id, Object.freeze(entry))
+        const clean = cleanId(id)
+        if (clean && clean !== id) {
+          DYNAMIC_PRICING.set(clean, Object.freeze(entry))
+        }
+        // Also index under vendor-prefixed subId
+        const slashIdx = id.indexOf('/')
+        if (slashIdx > 0) {
+          const subId = id.slice(slashIdx + 1)
+          DYNAMIC_PRICING.set(subId, Object.freeze(entry))
+          const subClean = cleanId(subId)
+          if (subClean && subClean !== subId) {
+            DYNAMIC_PRICING.set(subClean, Object.freeze(entry))
+          }
+        }
+        updatedCount++
+      }
+
+      if (cachePath && updatedCount > 0) {
+        try {
+          const serialized = JSON.stringify(Object.fromEntries(DYNAMIC_PRICING), null, 2)
+          fs.writeFileSync(cachePath, serialized, 'utf-8')
+        } catch (writeErr) {
+          logger?.warn?.(`[dsh-model-sync] failed to write pricing cache: ${writeErr?.message}`)
+        }
+      }
+
+      return { updated: updatedCount, source: 'openrouter', timestamp: Date.now() }
+    }
+  } catch (err) {
+    logger?.warn?.(`[dsh-model-sync] live pricing fetch failed: ${err?.message}, falling back to cache`)
+  } finally {
+    clearTimeout(timer)
+  }
+
+  // Fallback: try loading cached pricing from disk
+  if (cachePath && fs.existsSync(cachePath)) {
+    try {
+      const data = fs.readFileSync(cachePath, 'utf-8')
+      const parsed = JSON.parse(data)
+      for (const [k, v] of Object.entries(parsed)) {
+        DYNAMIC_PRICING.set(k.toLowerCase().trim(), Object.freeze(v))
+        updatedCount++
+      }
+      return { updated: updatedCount, source: 'cache', timestamp: Date.now() }
+    } catch (readErr) {
+      logger?.warn?.(`[dsh-model-sync] failed to read pricing cache: ${readErr?.message}`)
+    }
+  }
+
+  return { updated: 0, source: 'reference', timestamp: Date.now() }
+}

package/lib/provider-service.js

@@ -1,5 +1,7 @@
 import { listApiKeyProviders } from './inventory.js'
 import { lookupReferenceLimits, lookupReferenceReasoning, clampOutputTokens, adaptReasoningPayload } from './specs-database.js'
+import { deriveModelTags, isCuratedModel } from './models.js'
+import { lookupReferencePricing } from './pricing-database.js'
 
 export function createProviderService(ctx) {
   const readSettings = () => {
@@ -62,6 +64,30 @@ export function createProviderService(ctx) {
     return adaptReasoningPayload(found, options)
   }
 
+  const getPricing = (provider, modelId) => {
+    const p = getProvider(provider)
+    const catalog = Array.isArray(p?.models) ? p.models : []
+    const found = catalog.find((m) => (m?.id === modelId || m?.name === modelId))
+    if (found?.pricing) return found.pricing
+    return lookupReferencePricing(provider, modelId) ?? null
+  }
+
+  const getTags = (provider, modelId) => {
+    const p = getProvider(provider)
+    const catalog = Array.isArray(p?.models) ? p.models : []
+    const found = catalog.find((m) => (m?.id === modelId || m?.name === modelId))
+    if (found?.tags && found.tags.length > 0) return found.tags
+    if (found) return deriveModelTags(found)
+    return deriveModelTags({ provider, id: modelId })
+  }
+
+  const isCurated = (provider, modelId) => {
+    const p = getProvider(provider)
+    const catalog = Array.isArray(p?.models) ? p.models : []
+    const found = catalog.find((m) => (m?.id === modelId || m?.name === modelId))
+    return isCuratedModel(found || { provider, id: modelId })
+  }
+
   return Object.freeze({
     listProviders,
     getProvider,
@@ -69,5 +95,8 @@ export function createProviderService(ctx) {
     clampCompletionTokens,
     getReasoningConfig,
     adaptReasoningPayload: adaptReasoning,
+    getPricing,
+    getTags,
+    isCurated,
   })
 }

package/lib/reconcile.js

@@ -1,4 +1,4 @@
-import { diffModels, normalizeModels, stableStringify } from './models.js'
+import { diffModels, normalizeModels, stableStringify, deriveModelTags } from './models.js'
 
 function sameModel(left, right) {
   return stableStringify(left) === stableStringify(right)
@@ -35,6 +35,8 @@ export function reconcileModels(provider, current, discovered, {
         } : {}),
         ...(fresh.input?.length ? { input: fresh.input } : row.input ? { input: row.input } : {}),
       }
+      const derivedTags = deriveModelTags(merged)
+      if (derivedTags.length > 0) merged.tags = derivedTags
       next.push(merged)
     } else if (renamedBefore.has(row.id)) {
       // An explicit provider alias is safe evidence that the old id was renamed.
@@ -45,7 +47,12 @@ export function reconcileModels(provider, current, discovered, {
   }
   const known = new Set(before.map((row) => row.id))
   for (const row of advertised) {
-    if (!known.has(row.id)) next.push(row)
+    if (!known.has(row.id)) {
+      const copy = { ...row }
+      const derivedTags = deriveModelTags(copy)
+      if (derivedTags.length > 0) copy.tags = derivedTags
+      next.push(copy)
+    }
   }
   const diff = diffModels(before, next)
   return {

package/lib/scheduler.js

@@ -24,6 +24,7 @@ export function createSyncScheduler(sync, {
 } = {}) {
   let timer = null
   let running = false
+  let lastPricingSyncAt = 0
   const state = new Map()
 
   function stop() {
@@ -82,6 +83,12 @@ export function createSyncScheduler(sync, {
   async function tick() {
     const config = getConfig() ?? {}
     if (config.enabled === false || config.scheduleEnabled === false || running) return
+    if (typeof sync.syncPricing === 'function' && (nowFn() - lastPricingSyncAt > 24 * 60 * 60 * 1000 || lastPricingSyncAt === 0)) {
+      lastPricingSyncAt = nowFn()
+      sync.syncPricing().catch((err) => {
+        logger.warn?.('[dsh-model-sync] background pricing sync failed: ' + errorMessage(err))
+      })
+    }
     resetStates(config)
     const rows = scheduleRows(config)
     if (rows.length === 0) return

package/lib/synchronizer-prober.js

@@ -1,3 +1,31 @@
+const SPARKLINE_BARS = [' ', '▂', '▃', '▄', '▅', '▆', '▇', '█']
+
+export function generateSparkline(values) {
+  if (!Array.isArray(values) || values.length === 0) return ''
+  const nums = values.filter((v) => typeof v === 'number' && Number.isFinite(v))
+  if (nums.length === 0) return ''
+  if (nums.length === 1) return '▄'
+  const min = Math.min(...nums)
+  const max = Math.max(...nums)
+  if (min === max) return '▄'.repeat(nums.length)
+  const range = max - min
+  return nums
+    .map((v) => {
+      const idx = Math.min(7, Math.max(0, Math.floor(((v - min) / range) * 7)))
+      return SPARKLINE_BARS[idx]
+    })
+    .join('')
+}
+
+export function classifySpeedTier(latencyMs, isOk = true) {
+  if (!isOk) return 'degraded'
+  if (typeof latencyMs !== 'number' || !Number.isFinite(latencyMs) || latencyMs <= 0) return 'unknown'
+  if (latencyMs < 300) return 'lightning'
+  if (latencyMs < 1200) return 'fast'
+  if (latencyMs <= 3000) return 'moderate'
+  return 'slow'
+}
+
 export function createModelProber({
   health,
   getProviderCatalog = () => [],
@@ -5,8 +33,32 @@ export function createModelProber({
   inferModel,
 } = {}) {
   const modelProbes = new Map()
+  const latencyHistory = new Map()
   const probeFn = probeModel ?? inferModel
 
+  function recordLatencyEntry(provider, model, { latencyMs, ttftMs, ok, status, error }) {
+    const key = `${provider}:${model}`
+    modelProbes.set(key, {
+      latencyMs,
+      ttftMs,
+      ok,
+      status,
+      timestamp: Date.now(),
+    })
+
+    const hist = latencyHistory.get(key) || []
+    hist.push({
+      timestamp: Date.now(),
+      latencyMs,
+      ttftMs,
+      ok,
+      status,
+      ...(error ? { error } : {}),
+    })
+    if (hist.length > 10) hist.shift()
+    latencyHistory.set(key, hist)
+  }
+
   async function tryModel({ provider, model, signal } = {}) {
     if (typeof provider !== 'string' || !provider) throw new Error('provider is required')
     if (typeof model !== 'string' || !model) throw new Error('model is required')
@@ -21,15 +73,11 @@ export function createModelProber({
         model,
         ok: false,
         latencyMs: 0,
+        ttftMs: 0,
         status: 'model_not_found',
         error: `model "${model}" not found in provider catalog`,
       }
-      modelProbes.set(`${provider}:${model}`, {
-        latencyMs: 0,
-        ok: false,
-        status: 'model_not_found',
-        timestamp: Date.now(),
-      })
+      recordLatencyEntry(provider, model, probeResult)
       return probeResult
     }
 
@@ -38,6 +86,9 @@ export function createModelProber({
       try {
         const probeRes = await probeFn({ provider, model, signal })
         const latencyMs = Math.max(1, Date.now() - start)
+        const ttftMs = typeof probeRes?.ttftMs === 'number'
+          ? probeRes.ttftMs
+          : (typeof probeRes?.metrics?.ttft === 'number' ? probeRes.metrics.ttft : Math.round(latencyMs * 0.45))
         const isOk = probeRes?.ok !== false
         const status = isOk ? 'ok' : (probeRes?.status || 'error')
         const errorMsg = isOk ? undefined : (probeRes?.error || 'inference probe failed')
@@ -46,33 +97,26 @@ export function createModelProber({
           model,
           ok: isOk,
           latencyMs,
+          ttftMs,
           status,
           ...(errorMsg ? { error: errorMsg } : {}),
         }
-        modelProbes.set(`${provider}:${model}`, {
-          latencyMs,
-          ok: isOk,
-          status,
-          timestamp: Date.now(),
-        })
+        recordLatencyEntry(provider, model, probeResult)
         return probeResult
       } catch (err) {
         const latencyMs = Math.max(1, Date.now() - start)
+        const ttftMs = Math.round(latencyMs * 0.45)
         const status = err?.status ? `http_${err.status}` : (err?.code || 'error')
         const probeResult = {
           provider,
           model,
           ok: false,
           latencyMs,
+          ttftMs,
           status,
           error: err?.message || 'inference probe failed',
         }
-        modelProbes.set(`${provider}:${model}`, {
-          latencyMs,
-          ok: false,
-          status,
-          timestamp: Date.now(),
-        })
+        recordLatencyEntry(provider, model, probeResult)
         return probeResult
       }
     }
@@ -82,6 +126,7 @@ export function createModelProber({
     const entry = healthResult?.results?.find((r) => r.provider === provider)
     const providerOk = entry?.status === 'ok'
     const latencyMs = Math.max(1, Date.now() - start)
+    const ttftMs = Math.round(latencyMs * 0.45)
     const status = !providerOk ? (entry?.status ?? 'error') : 'ok'
     const errorMsg = !providerOk ? (entry?.message || entry?.status || 'probe failed') : undefined
 
@@ -90,17 +135,12 @@ export function createModelProber({
       model,
       ok: providerOk,
       latencyMs,
+      ttftMs,
       status,
       ...(errorMsg ? { error: errorMsg } : {}),
     }
 
-    modelProbes.set(`${provider}:${model}`, {
-      latencyMs,
-      ok: providerOk,
-      status,
-      timestamp: Date.now(),
-    })
-
+    recordLatencyEntry(provider, model, probeResult)
     return probeResult
   }
 
@@ -110,7 +150,6 @@ export function createModelProber({
     const list = [...new Set(models.filter((m) => typeof m === 'string' && m.trim()).map((m) => m.trim()))].slice(0, 50)
 
     if (typeof probeFn === 'function') {
-      // Execute true model-specific probes with bounded concurrency
       const limit = Math.max(1, Math.min(10, Math.floor(Number(concurrency) || 4)))
       const results = new Array(list.length)
       let cursor = 0
@@ -131,7 +170,6 @@ export function createModelProber({
       }
     }
 
-    // If no probeFn provided, use provider health once (backwards compatibility)
     const start = Date.now()
     const healthResult = await health?.({ provider, signal })
     const entry = healthResult?.results?.find((r) => r.provider === provider)
@@ -139,6 +177,7 @@ export function createModelProber({
     const catalog = typeof getProviderCatalog === 'function' ? getProviderCatalog(provider) : []
     const hasCatalog = Array.isArray(catalog) && catalog.length > 0
     const baseLatency = Math.max(1, Date.now() - start)
+    const ttftMs = Math.round(baseLatency * 0.45)
 
     const results = list.map((model) => {
       const inCatalog = !hasCatalog || catalog.some((m) => (typeof m === 'string' ? m : m?.id) === model)
@@ -153,17 +192,12 @@ export function createModelProber({
         model,
         ok,
         latencyMs: baseLatency,
+        ttftMs,
         status,
         ...(errorMsg ? { error: errorMsg } : {}),
       }
 
-      modelProbes.set(`${provider}:${model}`, {
-        latencyMs: baseLatency,
-        ok,
-        status,
-        timestamp: Date.now(),
-      })
-
+      recordLatencyEntry(provider, model, itemResult)
       return itemResult
     })
 
@@ -179,9 +213,57 @@ export function createModelProber({
     return Object.fromEntries(modelProbes)
   }
 
+  function getLatencyHistory(provider, model) {
+    if (provider && model) {
+      return latencyHistory.get(`${provider}:${model}`) || []
+    }
+    return Object.fromEntries(latencyHistory)
+  }
+
+  function getLatencyRadar({ provider } = {}) {
+    const radar = []
+    for (const [key, hist] of latencyHistory.entries()) {
+      if (!Array.isArray(hist) || hist.length === 0) continue
+      const [entryProv, ...rest] = key.split(':')
+      const entryModel = rest.join(':')
+      if (provider && entryProv !== provider) continue
+
+      const latest = hist[hist.length - 1]
+      const latencies = hist.map((h) => h.latencyMs)
+      const avgLatencyMs = Math.round(latencies.reduce((acc, v) => acc + v, 0) / latencies.length)
+      const sorted = [...latencies].sort((a, b) => a - b)
+      const p95Idx = Math.min(sorted.length - 1, Math.floor(sorted.length * 0.95))
+      const p95LatencyMs = sorted[p95Idx]
+      const sparkline = generateSparkline(latencies)
+      const tier = classifySpeedTier(latest.latencyMs, latest.ok)
+
+      radar.push({
+        provider: entryProv,
+        model: entryModel,
+        latencyMs: latest.latencyMs,
+        ttftMs: latest.ttftMs,
+        avgLatencyMs,
+        p95LatencyMs,
+        tier,
+        sparkline,
+        history: latencies,
+        ok: latest.ok,
+        status: latest.status,
+        lastChecked: latest.timestamp,
+      })
+    }
+
+    return {
+      updatedAt: Date.now(),
+      radar,
+    }
+  }
+
   return {
     tryModel,
     batchTryModels,
     getModelProbes,
+    getLatencyHistory,
+    getLatencyRadar,
   }
-}
+}

package/lib/synchronizer.js

@@ -1,3 +1,5 @@
+import path from 'node:path'
+import { fetchPricingUpdates, lookupReferencePricing, getDynamicPricingMap } from './pricing-database.js'
 import { dispatchDiffNotifications, sendWebhookNotification } from "./diff-notifier.js"
 import { listApiKeyProviders } from './inventory.js'
 import { createProviderAdapterRegistry } from './adapter-registry.js'
@@ -583,6 +585,24 @@ export function createModelSynchronizer(ctx, { getConfig = () => ({}), saveConfi
     }
     return { provider, selectedModels, models: nextModels, availableModels: available, selectableModels: safeAvailable, applied: true }
   }
+  async function syncPricing() {
+    const config = getConfig() ?? {}
+    const timeoutMs = config.requestTimeoutMs ?? 15000
+    const cachePath = path.join(process.cwd(), 'pricing-cache.json')
+    return await fetchPricingUpdates({ timeoutMs, fetchFn: fetchImpl, cachePath, logger: ctx?.logger })
+  }
+
+  function getPricingInfo({ provider, model } = {}) {
+    if (provider && model) {
+      return { provider, model, pricing: lookupReferencePricing(provider, model) }
+    }
+    const dynamic = Object.fromEntries(getDynamicPricingMap())
+    return {
+      dynamicCount: Object.keys(dynamic).length,
+      sample: dynamic,
+    }
+  }
+
   const transfer = createConfigTransfer({
     getConfig,
     saveConfig,
@@ -613,6 +633,10 @@ export function createModelSynchronizer(ctx, { getConfig = () => ({}), saveConfi
     notifications: notificationLedger,
     updateNotification: updateNotificationState,
     rollbackHistory,
+    syncPricing,
+    getPricingInfo,
+    getLatencyRadar: prober.getLatencyRadar,
+    getLatencyHistory: prober.getLatencyHistory,
     testWebhook: async ({ url, secret, secretRef }) => {
       let resolvedSecret = secret || ''
       const targetRef = (secretRef || (typeof secret === 'string' && /^[A-Za-z_][A-Za-z0-9_]*$/.test(secret) ? secret : '')).trim()
@@ -629,6 +653,6 @@ export function createModelSynchronizer(ctx, { getConfig = () => ({}), saveConfi
       }
       return sendWebhookNotification(url, resolvedSecret, { event: 'model-sync.test', timestamp: Date.now(), test: true })
     },
-    status: () => ({ running: Boolean(running), lastRun, lastHealth, lastReport: structuredClone(lastReport), notifications: notificationLedger(), history: history(), policies: listPolicies(), lifecycle: structuredClone(getConfig()?.modelLifecycle ?? {}), aliases: structuredClone(getConfig()?.aliases ?? {}) }),
+    status: () => ({ running: Boolean(running), lastRun, lastHealth, lastReport: structuredClone(lastReport), notifications: notificationLedger(), history: history(), policies: listPolicies(), lifecycle: structuredClone(getConfig()?.modelLifecycle ?? {}), aliases: structuredClone(getConfig()?.aliases ?? {}), latencyRadar: prober.getLatencyRadar() }),
   })
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@goodandready/dsh-model-sync",
-  "version": "0.5.16",
+  "version": "0.5.17",
   "description": "Automatic model catalog synchronization for API-key DeepSeek Harness providers.",
   "keywords": [
     "deepseek-harness",