dsh-cost-meter 1.5.5

package/lib/pricing.js

@@ -0,0 +1,799 @@
+/**
+ * DeepSeek 官方定价模型:价格表、官方文档解析、计费数学。
+ *
+ * 价格单位:美元 / 1M tokens(与官方文档一致)。账本中的成本恒以美元存储,
+ * 币种/汇率仅是展示层换算(config.exchangeRate)。
+ *
+ * 官方页面(2026-08-15 抓取)要点:
+ *  - 现为纯峰谷两档计价:空闲时段(OFF-PEAK)价格 = 高峰时段(PEAK)价格的一半;
+ *    deepseek-v4-flash 空闲 命中 $0.007 / 未命中 $0.22 / 输出 $0.66,
+ *    高峰 命中 $0.014 / 未命中 $0.44 / 输出 $1.32;
+ *    deepseek-v4-pro 空闲 命中 $0.022 / 未命中 $0.66 / 输出 $1.98,
+ *    高峰 命中 $0.044 / 未命中 $1.32 / 输出 $3.96。
+ *  - 峰时段为 01:00-04:00 与 06:00-10:00 UTC,其余为空闲时段;
+ *  - 页面已不再列出基础价档与生效时间(两档方案即时生效);本插件把空闲档
+ *    同时作为「基础档」存储,未启用峰谷计价时按空闲档计费。
+ *  - 页面未单列 cache write 价格,历史定价中 cache write 按 cache hit 计,
+ *    本插件沿用该规则(cacheRead + cacheWrite 均按命中价计)。
+ *
+ * 价格表写法:
+ *  - 三桶:{ cacheHit, cacheMiss, output }(DeepSeek 官方结构);
+ *  - 两档简写:{ input, output }(Anthropic / Gemini / Mistral 等无缓存折扣模型);
+ *  - 任意子集皆可:cacheMiss 缺省取 input,cacheHit 缺省取 cacheMiss(无缓存折扣
+ *    时命中价 = 未命中价),output 缺省为 0;峰谷子档(offPeak/peak/legacyBase)
+ *    同样适用该补齐规则。
+ */
+
+/** 官方定价页(英文版,服务端预渲染,可解析)。 */
+export const OFFICIAL_PRICING_URL = 'https://api-docs.deepseek.com/quick_start/pricing'
+
+/** 峰谷计价生效时间(UTC)。两档方案已即时生效:置为过去时刻,门控恒通过。 */
+export const DEFAULT_PEAK_EFFECTIVE_AT = '2026-08-01T00:00:00Z'
+
+/** 峰谷时代分界(2026-08-16 16:00 UTC):此前的计费按当时的基础价执行(历史正确性)。 */
+export const LEGACY_BASE_BOUNDARY = '2026-08-16T16:00:00Z'
+
+/** 峰谷时代之前的官方基础价(美元 / 1M tokens),历史计费按此执行。 */
+export const LEGACY_BASE_PRICES = {
+  'deepseek-v4-flash': { cacheHit: 0.0028, cacheMiss: 0.14, output: 0.28 },
+  'deepseek-v4-pro': { cacheHit: 0.003625, cacheMiss: 0.435, output: 0.87 },
+  // 旧模型别名:其基础价即历史价(官方已下架)。
+  'deepseek-chat': { cacheHit: 0.07, cacheMiss: 0.27, output: 1.1 },
+  'deepseek-reasoner': { cacheHit: 0.14, cacheMiss: 0.55, output: 2.19 },
+}
+
+/** 峰时段窗口(UTC 小时,半开区间 [start, end))。 */
+export const DEFAULT_PEAK_WINDOWS = [
+  { start: 1, end: 4 },
+  { start: 6, end: 10 },
+]
+
+/** 首批人工核对的非 DeepSeek 官方 token 价格(USD / 1M tokens)。 */
+export const DEFAULT_PROVIDER_PRICE_TABLE = {
+  openai: {
+    models: {
+      'gpt-5.6-sol': { input: 5, cachedInput: 0.5, output: 30, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '≤272K 档;超过 272K 输入按 $10/$45 计;缓存写入 $6.25' },
+      'gpt-5.6-terra': { input: 2, cachedInput: 0.2, output: 12, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '≤272K 档;超过 272K 按 $4/$18 计;缓存写入 $2.50' },
+      'gpt-5.6-luna': { input: 0.2, cachedInput: 0.02, output: 1.2, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '≤272K 档;超过 272K 按 $0.40/$1.80 计;缓存写入 $0.25' },
+      'gpt-5.5': { input: 5, cachedInput: 0.5, output: 30, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '≤272K 档;超过 272K 按 $10/$45 计' },
+      'gpt-5.5-pro': { input: 30, output: 180, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'gpt-5.4': { input: 2.5, cachedInput: 0.25, output: 15, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '≤272K 档;超过 272K 按 $5/$22.5 计' },
+      'gpt-5.4-pro': { input: 30, output: 180, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'gpt-5.4-mini': { input: 0.75, cachedInput: 0.075, output: 4.5, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'gpt-5.4-nano': { input: 0.2, cachedInput: 0.02, output: 1.25, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'gpt-5.3-codex': { input: 1.75, cachedInput: 0.175, output: 14, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'gpt-5.3-codex-spark': { input: 1.75, cachedInput: 0.175, output: 14, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'gpt-5.2': { input: 1.75, cachedInput: 0.175, output: 14, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'gpt-5.2-codex': { input: 1.75, cachedInput: 0.175, output: 14, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '2026-07-23 起弃用' },
+      'gpt-5.1': { input: 1.07, cachedInput: 0.107, output: 8.5, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'gpt-5.1-codex': { input: 1.07, cachedInput: 0.107, output: 8.5, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '2026-07-23 起弃用' },
+      'gpt-5.1-codex-max': { input: 1.25, cachedInput: 0.125, output: 10, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '2026-07-23 起弃用' },
+      'gpt-5.1-codex-mini': { input: 0.25, cachedInput: 0.025, output: 2, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '2026-07-23 起弃用' },
+      'gpt-5': { input: 1.07, cachedInput: 0.107, output: 8.5, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'gpt-5-codex': { input: 1.07, cachedInput: 0.107, output: 8.5, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '2026-07-23 起弃用' },
+      'gpt-5-nano': { input: 0.05, cachedInput: 0.005, output: 0.4, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'gpt-5-2025-08-07': { input: 1.25, cachedInput: 0.13, output: 10, billingMode: 'flat', notes: '2025-08 首发价;当前同名 gpt-5 已调价,历史 id 保留' },
+      'gpt-4.1-2025-04-14': { input: 2, cachedInput: 0.5, output: 8, billingMode: 'flat' },
+      'gpt-4.1-mini-2025-04-14': { input: 0.4, cachedInput: 0.1, output: 1.6, billingMode: 'flat' },
+    },
+  },
+  anthropic: {
+    models: {
+      'claude-fable-5': { input: 10, cachedInput: 1, output: 50, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '缓存写入 $12.50' },
+      'claude-opus-5': { input: 5, cachedInput: 0.5, output: 25, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '缓存写入 $6.25' },
+      'claude-opus-4-8': { input: 5, cachedInput: 0.5, output: 25, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '缓存写入 $6.25' },
+      'claude-opus-4-7': { input: 5, cachedInput: 0.5, output: 25, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'claude-opus-4-6': { input: 5, cachedInput: 0.5, output: 25, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'claude-opus-4-5': { input: 5, cachedInput: 0.5, output: 25, billingMode: 'flat', notes: '缓存写入 $6.25' },
+      'claude-sonnet-5': { input: 2, cachedInput: 0.2, output: 10, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '早鸟价(至 2026-08-31);标准价 $3/$15' },
+      'claude-sonnet-4-6': { input: 3, cachedInput: 0.3, output: 15, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '缓存写入 $3.75' },
+      'claude-sonnet-4-5': { input: 3, cachedInput: 0.3, output: 15, billingMode: 'flat', notes: '≤200K 档;>200K 按 $6/$22.5 计;缓存写入 $3.75' },
+      'claude-haiku-4-5': { input: 1, cachedInput: 0.1, output: 5, billingMode: 'flat', notes: '缓存写入 $1.25' },
+    },
+  },
+  google: {
+    models: {
+      'gemini-3.7-flash': { input: 0.75, output: 3.75, billingMode: 'flat', sourceUrl: 'https://ai.google.dev/gemini-api/docs/pricing', checkedAt: '2026-08-17', notes: '2026 年底前促销价;2027 起 $1.5/$7.5;Go 网关按 $1.5/$7.5' },
+      'gemini-3.6-flash': { input: 1.5, cachedInput: 0.15, output: 7.5, billingMode: 'flat', sourceUrl: 'https://ai.google.dev/gemini-api/docs/pricing', checkedAt: '2026-08-17' },
+      'gemini-3.5-flash': { input: 1.5, cachedInput: 0.15, output: 9, billingMode: 'flat', sourceUrl: 'https://ai.google.dev/gemini-api/docs/pricing', checkedAt: '2026-08-17' },
+      'gemini-3.5-flash-lite': { input: 0.3, cachedInput: 0.03, output: 2.5, billingMode: 'flat', sourceUrl: 'https://ai.google.dev/gemini-api/docs/pricing', checkedAt: '2026-08-17' },
+      'gemini-3.1-pro-preview': { input: 2, cachedInput: 0.2, output: 12, billingMode: 'flat', sourceUrl: 'https://ai.google.dev/gemini-api/docs/pricing', checkedAt: '2026-08-17', notes: '≤200K 档;>200K 按 $4/$18 计' },
+      'gemini-3-flash': { input: 0.5, cachedInput: 0.05, output: 3, billingMode: 'flat', sourceUrl: 'https://ai.google.dev/gemini-api/docs/pricing', checkedAt: '2026-08-17', notes: 'Preview' },
+      'gemini-2.5-pro': { input: 1.25, cachedInput: 0.125, output: 10, billingMode: 'flat', notes: '≤200K 档;>200K 按 $2.5/$15 计' },
+      'gemini-2.5-flash': { input: 0.3, cachedInput: 0.03, output: 2.5, billingMode: 'flat', sourceUrl: 'https://ai.google.dev/gemini-api/docs/pricing', checkedAt: '2026-08-17' },
+      'gemini-2.5-flash-lite': { input: 0.1, cachedInput: 0.01, output: 0.4, billingMode: 'flat', sourceUrl: 'https://ai.google.dev/gemini-api/docs/pricing', checkedAt: '2026-08-17' },
+    },
+  },
+  moonshot: {
+    models: {
+      'kimi-k3': { input: 3, cachedInput: 0.3, output: 15, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '官方人民币价 ¥20/¥100/缓存¥2 每百万;此处为 Go 网关美元价' },
+      'kimi-k2.7-code': { input: 0.95, cachedInput: 0.19, output: 4, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'kimi-k2.6': { input: 0.95, cachedInput: 0.16, output: 4, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'kimi-k2.5': { input: 0.6, cachedInput: 0.1, output: 3, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '2026-08-05 起弃用' },
+    },
+  },
+  'z-ai': {
+    models: {
+      'glm-5.3': { unpriced: true, billingMode: 'flat', notes: 'OpenCode Go 目录在册;官方/Zen 均未公布单价,不编造价格' },
+      'glm-5.2': { input: 1.4, cachedInput: 0.26, output: 4.4, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'glm-5.1': { input: 1.4, cachedInput: 0.26, output: 4.4, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'glm-5': { input: 1, cachedInput: 0.2, output: 3.2, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '2026-05-14 起弃用' },
+    },
+  },
+  xai: {
+    models: {
+      'grok-4.6': { input: 2, cachedInput: 0.5, output: 6, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '≤200K 档;>200K 按 $4/$12 计' },
+      'grok-4.5': { input: 2, cachedInput: 0.3, output: 6, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '≤200K 档;>200K 按 $4/$12 计' },
+      'grok-4.3': { input: 1.25, output: 2.5, billingMode: 'flat', sourceUrl: 'https://docs.x.ai/developers/models/grok-4.3' },
+      'grok-build-0.1': { input: 1, cachedInput: 0.2, output: 2, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+    },
+  },
+  alibaba: {
+    models: {
+      'qwen3.8-max': { input: 2, cachedInput: 0.25, output: 6, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17', notes: '缓存写入 $2.50' },
+      'qwen3.7-max': { input: 2.5, cachedInput: 0.5, output: 7.5, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17', notes: '缓存写入 $3.125' },
+      'qwen3.7-plus': { input: 0.4, cachedInput: 0.04, output: 1.6, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17', notes: '≤256K 档;>256K 按 $1.2/$4.8 计;缓存写入 $0.50' },
+      'qwen3.6-plus': { input: 0.5, cachedInput: 0.05, output: 3, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17', notes: '≤256K 档;>256K 按 $2/$6 计;缓存写入 $0.625' },
+      'qwen3.5-plus': { input: 0.2, cachedInput: 0.02, output: 1.2, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '缓存写入 $0.25' },
+      'qwen3-plus': { input: 0.4, output: 1.2, billingMode: 'flat', sourceUrl: 'https://www.alibabacloud.com/help/en/model-studio/model-pricing' },
+    },
+  },
+  minimax: {
+    models: {
+      'minimax-m3': { input: 0.3, cachedInput: 0.06, output: 1.2, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'minimax-m2.7': { input: 0.3, cachedInput: 0.06, output: 1.2, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17' },
+      'minimax-m2.5': { input: 0.3, cachedInput: 0.06, output: 1.2, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/zen', checkedAt: '2026-08-17', notes: '2026-08-05 起弃用' },
+    },
+  },
+  tencent: {
+    models: {
+      'hunyuan-a13b': { input: 0.5, output: 2, billingMode: 'flat', notes: 'Official price is CNY; catalog value requires currency conversion before use.' },
+      'hy3': { input: 0.14, cachedInput: 0.035, output: 0.58, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17', notes: 'OpenCode Go 目录价;厂商归属未官宣' },
+    },
+  },
+  xiaomi: {
+    models: {
+      'mimo-v2.5': { input: 0.14, cachedInput: 0.0028, output: 0.28, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'mimo-v2.5-pro': { input: 0.435, cachedInput: 0.003625, output: 0.87, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+    },
+  },
+  upstage: {
+    models: {
+      'solar-pro4': { input: 0.3, cachedInput: 0.06, output: 1.2, billingMode: 'flat', sourceUrl: 'https://www.upstage.ai/pricing/api' },
+      'solar-pro3': { input: 0.15, cachedInput: 0.015, output: 0.6, billingMode: 'flat', sourceUrl: 'https://www.upstage.ai/pricing/api' },
+    },
+  },
+  nvidia: { models: { 'nvidia/nemotron-3-ultra-550b-a55b': { unpriced: true, billingMode: 'flat', notes: 'Official catalog does not publish token price.' } } },
+  mistral: {
+    models: {
+      'mistral-large-2512': { input: 0.5, cachedInput: 0.05, output: 1.5, billingMode: 'flat' },
+      'mistral-medium-3.5': { input: 1.5, cachedInput: 0.15, output: 7.5, billingMode: 'flat' },
+      'mistral-small-4.0': { input: 0.15, cachedInput: 0.015, output: 0.6, billingMode: 'flat' },
+    },
+  },
+  // OpenCode Go 订阅($10/月)包含的模型中非 DeepSeek 的 17 个:订阅制下请求不按 token 扣费,
+  // 此处为官方公布的参考单价(用于成本估算/对比),来源 opencode.ai/docs/go。
+  // DeepSeek V4 Flash/Pro 与官方主表重复,以官方为准(含峰谷两档),Go 目录不重复收录(v1.5.2 移除)。
+  'opencode-go': {
+    models: {
+      'grok-4.5': { input: 2, cachedInput: 0.3, output: 6, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'glm-5.3': { unpriced: true, billingMode: 'flat', notes: 'Go 目录在册;官方未公布单价' },
+      'glm-5.2': { input: 1.4, cachedInput: 0.26, output: 4.4, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'glm-5.1': { input: 1.4, cachedInput: 0.26, output: 4.4, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'gpt-5.6-luna': { input: 0.2, cachedInput: 0.02, output: 1.2, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'kimi-k3': { input: 3, cachedInput: 0.3, output: 15, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'kimi-k2.7-code': { input: 0.95, cachedInput: 0.19, output: 4, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'kimi-k2.6': { input: 0.95, cachedInput: 0.16, output: 4, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'mimo-v2.5': { input: 0.14, cachedInput: 0.0028, output: 0.28, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'mimo-v2.5-pro': { input: 0.435, cachedInput: 0.003625, output: 0.87, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'minimax-m3': { input: 0.3, cachedInput: 0.06, output: 1.2, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'minimax-m2.7': { input: 0.3, cachedInput: 0.06, output: 1.2, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'qwen3.8-max': { input: 2, cachedInput: 0.25, output: 6, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'qwen3.7-max': { input: 2.5, cachedInput: 0.5, output: 7.5, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'qwen3.7-plus': { input: 0.4, cachedInput: 0.04, output: 1.6, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'qwen3.6-plus': { input: 0.5, cachedInput: 0.05, output: 3, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+      'hy3': { input: 0.14, cachedInput: 0.035, output: 0.58, billingMode: 'flat', sourceUrl: 'https://opencode.ai/docs/go', checkedAt: '2026-08-17' },
+    },
+  },
+}
+
+/** 拓展价格表目录的模型家族分组(展示用;未列出的模型自成一家)。 */
+export const PROVIDER_MODEL_FAMILIES = {
+  deepseek: { 'deepseek-v4-flash': 'DeepSeek v4', 'deepseek-v4-pro': 'DeepSeek v4' },
+  openai: {
+    'gpt-5.6-sol': 'GPT-5.6', 'gpt-5.6-terra': 'GPT-5.6', 'gpt-5.6-luna': 'GPT-5.6',
+    'gpt-5.5': 'GPT-5.5', 'gpt-5.5-pro': 'GPT-5.5',
+    'gpt-5.4': 'GPT-5.4', 'gpt-5.4-pro': 'GPT-5.4', 'gpt-5.4-mini': 'GPT-5.4', 'gpt-5.4-nano': 'GPT-5.4',
+    'gpt-5.3-codex': 'GPT-5.3 Codex', 'gpt-5.3-codex-spark': 'GPT-5.3 Codex',
+    'gpt-5.2': 'GPT-5.2', 'gpt-5.2-codex': 'GPT-5.2',
+    'gpt-5.1': 'GPT-5.1', 'gpt-5.1-codex': 'GPT-5.1', 'gpt-5.1-codex-max': 'GPT-5.1', 'gpt-5.1-codex-mini': 'GPT-5.1',
+    'gpt-5': 'GPT-5', 'gpt-5-codex': 'GPT-5', 'gpt-5-nano': 'GPT-5', 'gpt-5-2025-08-07': 'GPT-5',
+    'gpt-4.1': 'GPT-4.1', 'gpt-4.1-mini': 'GPT-4.1',
+  },
+  anthropic: {
+    'claude-fable-5': 'Claude Fable',
+    'claude-opus-5': 'Claude Opus', 'claude-opus-4-8': 'Claude Opus', 'claude-opus-4-7': 'Claude Opus', 'claude-opus-4-6': 'Claude Opus', 'claude-opus-4-5': 'Claude 4.5',
+    'claude-sonnet-5': 'Claude Sonnet', 'claude-sonnet-4-6': 'Claude Sonnet', 'claude-sonnet-4-5': 'Claude 4.5',
+    'claude-haiku-4-5': 'Claude 4.5',
+  },
+  google: {
+    'gemini-3.7-flash': 'Gemini 3.7 Flash',
+    'gemini-3.6-flash': 'Gemini 3.6 Flash',
+    'gemini-3.5-flash': 'Gemini 3.5 Flash', 'gemini-3.5-flash-lite': 'Gemini 3.5 Flash',
+    'gemini-3.1-pro-preview': 'Gemini 3.1 Pro',
+    'gemini-3-flash': 'Gemini 3 Flash',
+    'gemini-2.5-pro': 'Gemini 2.5', 'gemini-2.5-flash': 'Gemini 2.5', 'gemini-2.5-flash-lite': 'Gemini 2.5',
+  },
+  moonshot: { 'kimi-k3': 'Kimi K3', 'kimi-k2.7-code': 'Kimi K2', 'kimi-k2.6': 'Kimi K2', 'kimi-k2.5': 'Kimi K2' },
+  'z-ai': { 'glm-5.3': 'GLM-5', 'glm-5.2': 'GLM-5', 'glm-5.1': 'GLM-5', 'glm-5': 'GLM-5' },
+  xai: { 'grok-4.6': 'Grok 4', 'grok-4.5': 'Grok 4', 'grok-4.3': 'Grok 4', 'grok-build-0.1': 'Grok Build' },
+  alibaba: { 'qwen3.8-max': 'Qwen3.8', 'qwen3.7-max': 'Qwen3.7', 'qwen3.7-plus': 'Qwen3.7', 'qwen3.6-plus': 'Qwen3.6', 'qwen3.5-plus': 'Qwen3.5', 'qwen3-plus': 'Qwen3' },
+  minimax: { 'minimax-m3': 'MiniMax M3', 'minimax-m2.7': 'MiniMax M2', 'minimax-m2.5': 'MiniMax M2' },
+  tencent: { 'hunyuan-a13b': '混元', 'hy3': 'Hy3' },
+  xiaomi: { 'mimo-v2.5': 'MiMo V2.5', 'mimo-v2.5-pro': 'MiMo V2.5' },
+  upstage: { 'solar-pro4': 'Solar', 'solar-pro3': 'Solar' },
+  nvidia: { 'nvidia/nemotron-3-ultra-550b-a55b': 'Nemotron' },
+  mistral: { 'mistral-large-2512': 'Mistral Large', 'mistral-medium-3.5': 'Mistral Medium', 'mistral-small-4.0': 'Mistral Small' },
+  'opencode-go': {
+    'gpt-5.6-luna': 'GPT', 'grok-4.5': 'Grok', 'glm-5.3': 'GLM', 'glm-5.2': 'GLM', 'glm-5.1': 'GLM',
+    'kimi-k3': 'Kimi', 'kimi-k2.7-code': 'Kimi', 'kimi-k2.6': 'Kimi',
+    'mimo-v2.5': 'MiMo', 'mimo-v2.5-pro': 'MiMo', 'minimax-m3': 'MiniMax', 'minimax-m2.7': 'MiniMax',
+    'qwen3.8-max': 'Qwen', 'qwen3.7-max': 'Qwen', 'qwen3.7-plus': 'Qwen', 'qwen3.6-plus': 'Qwen',
+    'hy3': 'Hy3',
+  },
+}
+
+/**
+ * 构建扩展价格表目录:provider → family → modelId → 价格条目。
+ * 内置只读目录(含 DeepSeek 当前模型);「挂载」= 把条目复制进可编辑价格表。
+ */
+export function buildPriceCatalog() {
+  const catalog = {}
+  const put = (provider, id, entry) => {
+    const family = PROVIDER_MODEL_FAMILIES[provider]?.[id] ?? id
+    if (catalog[provider] === undefined) catalog[provider] = {}
+    if (catalog[provider][family] === undefined) catalog[provider][family] = {}
+    catalog[provider][family][id] = entry
+  }
+  for (const [id, entry] of Object.entries(DEFAULT_PRICE_TABLE.models)) put('deepseek', id, entry)
+  for (const [provider, table] of Object.entries(DEFAULT_PROVIDER_PRICE_TABLE)) {
+    for (const [id, entry] of Object.entries(table.models)) put(provider, id, entry)
+  }
+  return catalog
+}
+
+/** 内置默认 DeepSeek 价格表(与官方页面当前数字一致,供首次启动使用;基础档 = 空闲档)。 */
+export const DEFAULT_PRICE_TABLE = {
+  models: {
+    'deepseek-v4-flash': {
+      cacheHit: 0.007,
+      cacheMiss: 0.22,
+      output: 0.66,
+      offPeak: { cacheHit: 0.007, cacheMiss: 0.22, output: 0.66 },
+      peak: { cacheHit: 0.014, cacheMiss: 0.44, output: 1.32 },
+      legacyBase: { cacheHit: 0.0028, cacheMiss: 0.14, output: 0.28 },
+    },
+    'deepseek-v4-pro': {
+      cacheHit: 0.022,
+      cacheMiss: 0.66,
+      output: 1.98,
+      offPeak: { cacheHit: 0.022, cacheMiss: 0.66, output: 1.98 },
+      peak: { cacheHit: 0.044, cacheMiss: 1.32, output: 3.96 },
+      legacyBase: { cacheHit: 0.003625, cacheMiss: 0.435, output: 0.87 },
+    },
+  },
+  default: { cacheHit: 0.007, cacheMiss: 0.22, output: 0.66 },
+}
+
+/**
+ * 补齐一档价格:支持多种模型计费写法。
+ *  - 三桶写法:{ cacheHit, cacheMiss, output }(DeepSeek 官方结构);
+ *  - 两档简写:{ input, output }(Anthropic / Gemini / Mistral 等无缓存折扣模型
+ *    的价表通常只给输入/输出两档);
+ *  - 混合:{ cacheMiss, output } 等任意子集。
+ * 补齐规则:
+ *  - cacheMiss 缺省 → 取 input;两者都缺 → 0;
+ *  - cacheHit 缺省 → 取 cacheMiss(无缓存折扣时命中价 = 未命中价);
+ *  - output 缺省 → 0。
+ * 显式给出的数字恒优先;非负有限数字才被接受。
+ * @param raw - 任意一档价格对象。
+ * @returns 补全后的三桶价格 { cacheHit, cacheMiss, output },或 undefined。
+ */
+function completeTier(raw) {
+  if (raw === null || typeof raw !== 'object' || Array.isArray(raw)) return undefined
+  const n = key => {
+    const v = raw[key]
+    return typeof v === 'number' && Number.isFinite(v) && v >= 0 ? v : undefined
+  }
+  const cacheMiss = n('cacheMiss') ?? n('input') ?? 0
+  const cacheHit = n('cacheHit') ?? n('cachedInput') ?? n('cacheRead') ?? cacheMiss
+  const output = n('output') ?? 0
+  const reasoning = n('reasoning')
+  return reasoning === undefined ? { cacheHit, cacheMiss, output } : { cacheHit, cacheMiss, output, reasoning }
+}
+
+/**
+ * 规范化一条价格记录:按 completeTier 补齐缺失字段,剥离未知字段。
+ * @param value - 任意解析结果。
+ * @returns 规范化后的价格记录,或 null。
+ */
+export function normalizePrice(value) {
+  if (value === null || typeof value !== 'object' || Array.isArray(value)) return null
+  if (value.unpriced === true) {
+    const entry = { unpriced: true }
+    for (const key of ['billingMode', 'sourceUrl', 'checkedAt', 'notes']) if (typeof value[key] === 'string') entry[key] = value[key]
+    return entry
+  }
+  if (!('cacheHit' in value) && !('cacheMiss' in value) && !('output' in value) && !('input' in value)) return null
+  const entry = completeTier(value)
+  if (value.legacy === true) entry.legacy = true
+  if (value.billingMode === 'flat' || value.billingMode === 'deepseek-peak' || value.billingMode === 'batch') entry.billingMode = value.billingMode
+  for (const key of ['sourceUrl', 'checkedAt', 'notes']) if (typeof value[key] === 'string') entry[key] = value[key]
+  const offPeak = completeTier(value.offPeak)
+  if (offPeak !== undefined) entry.offPeak = offPeak
+  const peak = completeTier(value.peak)
+  if (peak !== undefined) entry.peak = peak
+  const legacyBase = completeTier(value.legacyBase)
+  if (legacyBase !== undefined) entry.legacyBase = legacyBase
+  return entry
+}
+
+/** 全部价格为 0 的记录视为空记录。 */
+export function isZeroPrice(entry) {
+  return entry !== null && entry.cacheHit === 0 && entry.cacheMiss === 0 && entry.output === 0
+}
+
+/**
+ * 按模型 id 解析价格记录:精确匹配 → default 回退。
+ * @param modelId - 请求中的模型 id。
+ * @param table - { models, default } 价格表。
+ * @returns 价格记录。
+ */
+export function priceEntryFor(modelId, table) {
+  const models = table?.models ?? {}
+  if (typeof modelId === 'string' && modelId.length > 0) {
+    const exact = models[modelId]
+    if (exact !== undefined) return exact
+    // 别名匹配:deepseek-chat → 任何以 '-' 连接的相近 id 不再猜测,直接回退 default。
+  }
+  return table?.default ?? { cacheHit: 0, cacheMiss: 0, output: 0 }
+}
+
+// ── 模型名自动匹配(精确 → 手动覆盖 → 去后缀/前缀/家族相似) ───────────
+
+/**
+ * 模型名归一化:小写,去掉括号括起的附注(如 (go)),再只保留字母与数字。
+ * 大小写、空格、横杠、下划线、点号等差异全部忽略:'GPT-5.6 Luna (Go)' → 'gpt56luna'。
+ */
+export function canonModelId(id) {
+  return String(id ?? '').toLowerCase()
+    .replace(/\([^)]*\)/g, ' ')
+    .replace(/（[^）]*）/g, ' ')
+    .replace(/[^a-z0-9]+/g, '')
+}
+
+function stripIdDecor(id) {
+  let out = String(id).toLowerCase()
+  // 去掉日期后缀(-2026-01-01 / -20260101 / @2026-01-01)与带 v 的版本号后缀(-v2 / -v1.5)。
+  // 注意:不剥裸数字后缀——'glm-5.3' 的 '-5.3' 是模型名本体而非版本后缀,
+  // 剥掉会让 glm-5.3/glm-5.2 都退化成 'glm' 而互配,把订阅制模型记到同家族付费价上(issue #18)。
+  out = out.replace(/[-@]\d{4}-?\d{2}-?\d{2}$/, '')
+  out = out.replace(/[-@]v\d+(\.\d+)*$/, '')
+  return out
+}
+
+const tokensOf = id => stripIdDecor(id).split(/[-_./:]+/).filter(Boolean)
+
+/**
+ * 把请求模型 id 匹配到候选价格表 id。
+ * 顺序:精确 → 归一化等价(忽略大小写/空格/横杠/点号/括号附注) → 宽泛包含
+ * (请求名归一化后包含候选名即算命中,取最长候选) → 去日期/版本后缀精确
+ * → 候选前缀(取最长) → 家族 token 相似(≥2 个前缀 token 且最长者胜)。
+ * @param modelId - 请求模型 id。
+ * @param candidates - 候选 id 数组。
+ * @returns 命中的候选 id,或 null。
+ */
+export function matchModelId(modelId, candidates) {
+  if (typeof modelId !== 'string' || modelId.length === 0) return null
+  const list = Array.isArray(candidates) ? candidates.filter(c => typeof c === 'string' && c.length > 0) : []
+  if (list.length === 0) return null
+  const exact = list.find(c => c === modelId)
+  if (exact !== undefined) return exact
+  const canon = canonModelId(modelId)
+  if (canon.length === 0) return null
+  // 归一化后等价:'GPT-5.6 Luna' ≡ 'gpt-5.6-luna'。
+  const byCanon = list.find(c => canonModelId(c) === canon)
+  if (byCanon !== undefined) return byCanon
+  // 宽泛包含:'gpt5.6 luna(go)' 归一化后包含 'gpt56luna' 即命中;取最长候选,过短候选(≤3)防误配。
+  let containHit = null
+  let containLen = 0
+  for (const c of list) {
+    const cc = canonModelId(c)
+    if (cc.length < 4 || cc === canon) continue
+    if (canon.includes(cc) && cc.length > containLen) { containHit = c; containLen = cc.length }
+  }
+  if (containHit !== null) return containHit
+  const stripped = stripIdDecor(modelId)
+  const byStripped = list.find(c => stripIdDecor(c) === stripped)
+  if (byStripped !== undefined) return byStripped
+  // 前缀匹配:modelId(去饰后)以候选(去饰后)开头且紧接分隔符,取最长候选。
+  let prefixHit = null
+  for (const c of list) {
+    const cs = stripIdDecor(c)
+    if (cs.length === 0 || cs === stripped) continue
+    if (stripped.startsWith(cs) && /^[\-_./:]/.test(stripped.slice(cs.length))) {
+      if (prefixHit === null || stripIdDecor(prefixHit).length < cs.length) prefixHit = c
+    }
+  }
+  if (prefixHit !== null) return prefixHit
+  // 家族 token 相似:前缀公共 token ≥2,取公共最长者;同长取候选最短(更泛化的家族)。
+  const mt = tokensOf(modelId)
+  if (mt.length < 2) return null
+  let best = null
+  let bestLen = 0
+  for (const c of list) {
+    const ct = tokensOf(c)
+    let n = 0
+    while (n < mt.length && n < ct.length && mt[n] === ct[n]) n += 1
+    // 防跨版本误配(issue #18):分歧位置两侧都有数字/版本号 token(如 glm-5.3 vs glm-5.2)
+    // 视为不同模型拒绝匹配——订阅制/新版本模型不应落到同家族其它版本的付费单价。
+    if (n < mt.length && n < ct.length && /^\d+$/.test(mt[n]) && /^\d+$/.test(ct[n])) continue
+    if (n >= 2 && (n > bestLen || (n === bestLen && best !== null && c.length < best.length))) {
+      best = c
+      bestLen = n
+    }
+  }
+  return best
+}
+
+/**
+ * provider-aware 价格查找。provider 缺失时保持旧版 DeepSeek 行为；
+ * 已知非 DeepSeek provider 未配置模型时返回 null，避免误套 DeepSeek default。
+ * @param options - { mode: 'auto'|'exact', overrides: { 'provider:modelId': '目标' } };
+ *   overrides 目标可为同 provider 模型 id,或 'provider:modelId' 跨 provider 引用;
+ *   'deepseek:__default__' 表示回退 DeepSeek 默认价。
+ */
+export function providerPriceEntryFor(provider, modelId, prices, options) {
+  const rawProvider = typeof provider === 'string' ? provider.trim().toLowerCase() : ''
+  const normalized = rawProvider.startsWith('llm-') ? rawProvider.slice(4) : rawProvider
+  const mode = options?.mode === 'exact' ? 'exact' : 'auto'
+  const overrides = options?.overrides !== null && typeof options?.overrides === 'object' ? options.overrides : {}
+  // 手动覆盖优先于自动匹配。
+  let targetModel = modelId
+  let targetProvider = normalized
+  const overrideKey = (normalized === '' ? 'deepseek' : normalized) + ':' + modelId
+  const override = overrides[overrideKey]
+  if (typeof override === 'string' && override.length > 0) {
+    const sep = override.indexOf(':')
+    if (sep > 0 && override.slice(sep + 1).length > 0) {
+      targetProvider = override.slice(0, sep).trim().toLowerCase()
+      targetModel = override.slice(sep + 1)
+    } else {
+      targetModel = override
+    }
+    if (targetProvider === 'deepseek' && targetModel === '__default__') {
+      return { entry: prices?.default ?? { cacheHit: 0, cacheMiss: 0, output: 0 }, billingMode: 'deepseek-peak', priced: true }
+    }
+  }
+  if (targetProvider === '' || targetProvider === 'deepseek' || targetProvider.includes('deepseek')) {
+    const models = prices?.models ?? {}
+    const hit = models[targetModel] !== undefined
+      ? targetModel
+      : (mode === 'auto' ? matchModelId(targetModel, Object.keys(models)) : null)
+    if (hit !== null && hit !== undefined) {
+      return { entry: models[hit], billingMode: 'deepseek-peak', priced: true }
+    }
+    const entry = priceEntryFor(targetModel, prices)
+    return { entry, billingMode: 'deepseek-peak', priced: entry?.unpriced !== true }
+  }
+  const providerTable = prices?.providers?.[targetProvider]
+  const catalog = providerTable?.models ?? {}
+  let hit = catalog[targetModel] !== undefined ? targetModel : null
+  if (hit === null && mode === 'auto') hit = matchModelId(targetModel, Object.keys(catalog))
+  if (hit === null && mode === 'auto') {
+    // 跨厂商兑底:请求携带的 provider 未在价格表登记(opencode / zen 等路由入口)时,
+    // 按模型名全库查找——先查 DeepSeek 主表(保留峰谷两档),再取其余厂商中归一化最长命中。
+    const dsModels = prices?.models ?? {}
+    const dsHit = matchModelId(targetModel, Object.keys(dsModels))
+    if (dsHit !== null) return { entry: dsModels[dsHit], billingMode: 'deepseek-peak', priced: true }
+    let bestEntry = null
+    let bestLen = 0
+    let bestMode = 'flat'
+    for (const [prov, table] of Object.entries(prices?.providers ?? {})) {
+      if (prov === targetProvider) continue
+      const models = table?.models ?? {}
+      const h = matchModelId(targetModel, Object.keys(models))
+      if (h === null) continue
+      const entry = normalizePrice(models[h])
+      if (entry === null || entry.unpriced === true) continue
+      const score = canonModelId(h).length
+      if (score > bestLen) {
+        bestEntry = entry
+        bestLen = score
+        bestMode = models[h]?.billingMode === 'deepseek-peak' ? 'deepseek-peak' : 'flat'
+      }
+    }
+    if (bestEntry !== null) return { entry: bestEntry, billingMode: bestMode, priced: true }
+  }
+  if (hit === null) return { entry: null, billingMode: 'flat', priced: false }
+  const entry = normalizePrice(catalog[hit])
+  if (entry === null || entry.unpriced === true) return { entry: null, billingMode: 'flat', priced: false }
+  return { entry, billingMode: catalog[hit].billingMode === 'deepseek-peak' ? 'deepseek-peak' : 'flat', priced: true }
+}
+
+/**
+ * 某一时刻是否处于峰时段。
+ * @param atMs - 时刻(epoch ms)。
+ * @param effectiveAtMs - 峰谷计价生效时刻(epoch ms)。
+ * @param windows - 峰时段窗口数组({start,end} UTC 小时,半开区间)。
+ * @returns 峰时段返回 true;生效前或窗口外返回 false。
+ */
+export function isPeakHour(atMs, effectiveAtMs, windows) {
+  if (!Array.isArray(windows) || windows.length === 0) return false
+  if (Number.isFinite(effectiveAtMs) && atMs < effectiveAtMs) return false
+  const hour = new Date(atMs).getUTCHours()
+  return windows.some(w => {
+    const start = Number(w?.start)
+    const end = Number(w?.end)
+    if (!Number.isFinite(start) || !Number.isFinite(end)) return false
+    if (start < end) return hour >= start && hour < end
+    // 跨午夜窗口(本配置不会出现,兼容处理)。
+    return hour >= start || hour < end
+  })
+}
+
+/**
+ * 某一时刻所处的峰谷相位与相邻相位切换点(供倒计时/进度条展示)。
+ * 窗口为半开区间 [start, end)(UTC 小时),兼容跨午夜窗口(end <= start)。
+ * @param atMs - 时刻(epoch ms)。
+ * @param windows - 峰时段窗口数组。
+ * @returns { inPeak, prevAtMs, nextAtMs, nextIntoPeak },或 null(无有效窗口/时刻)。
+ *   prevAtMs = 当前相位起点,nextAtMs = 下一次切换时刻,nextIntoPeak = 该次切换是否进入峰时段。
+ */
+export function peakPhaseAt(atMs, windows) {
+  if (!Array.isArray(windows) || windows.length === 0 || !Number.isFinite(atMs)) return null
+  const hourAt = (dayOffset, hour) => {
+    const date = new Date(atMs)
+    date.setUTCDate(date.getUTCDate() + dayOffset)
+    date.setUTCHours(hour, 0, 0, 0)
+    return date.getTime()
+  }
+  // 收集前一天到后一天的全部切换点,保证任意时刻都能取到前后相邻切换点。
+  const points = []
+  for (let day = -1; day <= 1; day += 1) {
+    for (const w of windows) {
+      const start = Number(w?.start)
+      const end = Number(w?.end)
+      if (!Number.isFinite(start) || !Number.isFinite(end)) continue
+      points.push({ at: hourAt(day, start), intoPeak: true })
+      // 跨午夜窗口的结束点落在次日。
+      points.push({ at: hourAt(end <= start ? day + 1 : day, end), intoPeak: false })
+    }
+  }
+  const inPeak = isPeakHour(atMs, undefined, windows)
+  let prev = null
+  let next = null
+  for (const p of points) {
+    if (p.at <= atMs && (prev === null || p.at > prev.at)) prev = p
+    if (p.at > atMs && (next === null || p.at < next.at)) next = p
+  }
+  if (prev === null || next === null) return null
+  return { inPeak, prevAtMs: prev.at, nextAtMs: next.at, nextIntoPeak: next.intoPeak }
+}
+
+/**
+ * 为一次用量挑选价格档位:生效后峰时段 → peak;生效后谷时段 → offPeak;
+ * 生效前(或禁用峰谷)→ 基础价格。cache write 与 cache hit 同价。
+ * @param entry - 模型价格记录。
+ * @param atMs - 计费时刻。
+ * @param peak - { enabled, effectiveAtMs, windows } 峰谷配置。
+ * @returns 三档价格 { cacheHit, cacheMiss, output }。
+ */
+export function tierFor(entry, atMs, peak) {
+  const base = entry ?? { cacheHit: 0, cacheMiss: 0, output: 0 }
+  const asTier = price => price.reasoning === undefined
+    ? { cacheHit: price.cacheHit, cacheMiss: price.cacheMiss, output: price.output }
+    : { cacheHit: price.cacheHit, cacheMiss: price.cacheMiss, output: price.output, reasoning: price.reasoning }
+  // 峰谷时代之前(2026-08-16 16:00 UTC 前):按当时的基础价计费(历史正确性)。
+  if (Number.isFinite(atMs) && atMs < Date.parse(LEGACY_BASE_BOUNDARY)) {
+    const lb = base.legacyBase
+    return lb === undefined ? asTier(base) : asTier(lb)
+  }
+  if (peak?.enabled !== true) return asTier(base)
+  const effectiveAtMs = typeof peak.effectiveAtMs === 'number' ? peak.effectiveAtMs : undefined
+  if (isPeakHour(atMs, effectiveAtMs, peak.windows)) {
+    const p = base.peak
+    return p === undefined ? asTier(base) : asTier(p)
+  }
+  if (effectiveAtMs !== undefined && atMs >= effectiveAtMs) {
+    const off = base.offPeak
+    return off === undefined ? asTier(base) : asTier(off)
+  }
+  return asTier(base)
+}
+
+/**
+ * 一次调用的美元成本。
+ * @param tokens - { input, output, cacheRead, cacheWrite } 各桶 token 数。
+ * @param entry - 模型价格记录。
+ * @param atMs - 计费时刻。
+ * @param peak - 峰谷配置。
+ * @returns 美元成本(非负)。
+ */
+export function costOf(tokens, entry, atMs, peak) {
+  const tier = tierFor(entry, atMs, peak)
+  const input = Math.max(0, Number(tokens?.input) || 0)
+  const output = Math.max(0, Number(tokens?.output) || 0)
+  const cacheRead = Math.max(0, Number(tokens?.cacheRead) || 0)
+  const cacheWrite = Math.max(0, Number(tokens?.cacheWrite) || 0)
+  const reasoning = Math.max(0, Number(tokens?.reasoning) || 0)
+  const reasoningPrice = typeof tier.reasoning === 'number' ? tier.reasoning : 0
+  const cost = (input * tier.cacheMiss
+    + output * tier.output
+    + (cacheRead + cacheWrite) * tier.cacheHit
+    + reasoning * reasoningPrice) / 1_000_000
+  return Math.max(0, cost)
+}
+
+/** 金额显示:美元成本 × 汇率,按币种格式化,截断而非四舍五入进位。 */
+export function formatMoney(usdCost, display) {
+  const rate = Number(display?.exchangeRate)
+  const value = usdCost * (Number.isFinite(rate) && rate > 0 ? rate : 1)
+  const symbol = typeof display?.symbol === 'string' && display.symbol.length > 0 ? display.symbol : '$'
+  const decimals = Math.max(0, Math.min(10, Math.floor(Number(display?.decimals) || 2)))
+  // 数值过小时自动放宽小数位,避免显示成 0。
+  let effective = decimals
+  if (value > 0 && value < 10 ** -decimals) effective = decimals + 2
+  const fixed = value.toFixed(effective)
+  const trimmed = fixed.includes('.') ? fixed.replace(/0+$/, '').replace(/\.$/, '') : fixed
+  return `${symbol}${trimmed}`
+}
+
+// ── 官方页面解析 ──────────────────────────────────────────────────────────
+
+function decodeEntities(text) {
+  return text
+    .replace(/&nbsp;/g, ' ')
+    .replace(/&amp;/g, '&')
+    .replace(/&lt;/g, '<')
+    .replace(/&gt;/g, '>')
+    .replace(/&quot;/g, '"')
+    .replace(/&#39;/g, "'")
+    .replace(/&mdash;/g, '—')
+}
+
+function stripTags(html) {
+  return decodeEntities(String(html).replace(/<[^>]*>/g, ' ')).replace(/\s+/g, ' ').trim()
+}
+
+/** 取出页面内所有 <table> 块,解析为行 × 单元格文本。 */
+function parseTables(html) {
+  const blocks = String(html).match(/<table[\s\S]*?<\/table>/gi) ?? []
+  return blocks.map(block => {
+    const rows = []
+    const trs = block.match(/<tr[\s\S]*?<\/tr>/gi) ?? []
+    for (const tr of trs) {
+      const cells = tr.match(/<t[dh][^>]*>([\s\S]*?)<\/t[dh]>/gi) ?? []
+      const row = cells.map(cell => stripTags(cell.replace(/^<t[dh][^>]*>/, '').replace(/<\/t[dh]>$/, '')))
+      if (row.length > 0) rows.push(row)
+    }
+    return rows
+  })
+}
+
+/** 单元格内的美元金额,取第一个 $ 数字。 */
+function cellMoney(cell) {
+  const m = /(?:^|\s)\$([0-9]+(?:\.[0-9]+)?)/.exec(cell ?? '')
+  if (m === null) return null
+  const value = Number(m[1])
+  return Number.isFinite(value) ? value : null
+}
+
+const MODEL_ID = /deepseek-[a-z0-9_.-]+/i
+
+/**
+ * 解析官方定价页 HTML。
+ *
+ * 页面为一张表(服务端预渲染,结构与 2026-08-15 抓取一致):
+ *  - 首行 [MODEL, <模型id>...] 给出全部模型 id;
+ *  - 计价行按指标分组:指标标签行 [1M INPUT TOKENS (CACHE HIT), OFF-PEAK, $hit, $hit]
+ *    后跟 PEAK 续行 [PEAK, $hit, $hit](首两格被上一行 rowspan 合并);
+ *  - 每个指标给出 OFF-PEAK / PEAK 两档各模型价格,空闲档 = 高峰档的一半;
+ *  - 页面已不再列出基础价档与生效时间(两档方案即时生效),因此 models 的
+ *    基础档直接取空闲档数值,effectiveAt 返回 null。
+ * @param html - 页面源文本。
+ * @returns { models, effectiveAt, peakWindows } 解析结果。
+ * @throws 无法识别价格表时抛出带说明的 Error。
+ */
+export function parsePricingHtml(html) {
+  const tables = parseTables(html)
+  const modelIds = []
+  /** metricKey -> { offPeak: number[], peak: number[] }(按模型顺序)。 */
+  const tiers = {}
+  const metricOf = cell => {
+    const text = (cell ?? '').trim().toUpperCase()
+    if (text.includes('CACHE HIT')) return 'cacheHit'
+    if (text.includes('CACHE MISS')) return 'cacheMiss'
+    if (text.includes('OUTPUT TOKENS')) return 'output'
+    return null
+  }
+
+  for (const rows of tables) {
+    let lastMetric = null
+    for (let i = 0; i < rows.length; i += 1) {
+      const row = rows[i]
+      const first = (row[0] ?? '').trim()
+      // 模型表头行:MODEL 后跟全部模型 id。
+      if (/^MODEL$/i.test(first)) {
+        const ids = row.slice(1).map(cell => (MODEL_ID.exec(cell ?? '') ?? [])[0]).filter(Boolean)
+        if (ids.length > 0) modelIds.splice(0, modelIds.length, ...ids)
+        continue
+      }
+      // 指标标签可能在本行任意单元格(含 rowspan 合并布局);PEAK 续行沿用上一行指标。
+      const metric = metricOf(row.join(' ')) ?? lastMetric
+      if (metric !== null) lastMetric = metric
+      // 档位标签:OFF-PEAK / PEAK,价格紧跟其后。
+      const tierIdx = row.findIndex(cell => /^OFF-PEAK$/i.test((cell ?? '').trim()) || /^PEAK$/i.test((cell ?? '').trim()))
+      if (tierIdx < 0) continue
+      if (metric === null || modelIds.length === 0) continue
+      const label = /^PEAK$/i.test((row[tierIdx] ?? '').trim()) ? 'peak' : 'offPeak'
+      const prices = row.slice(tierIdx + 1, tierIdx + 1 + modelIds.length).map(cellMoney)
+      if (prices.some(v => v === null)) continue
+      if (tiers[metric] === undefined) tiers[metric] = { offPeak: [], peak: [] }
+      tiers[metric][label] = prices
+    }
+  }
+
+  const models = {}
+  for (let k = 0; k < modelIds.length; k += 1) {
+    const id = modelIds[k].toLowerCase()
+    const off = {
+      cacheHit: tiers.cacheHit?.offPeak?.[k],
+      cacheMiss: tiers.cacheMiss?.offPeak?.[k],
+      output: tiers.output?.offPeak?.[k],
+    }
+    const pk = {
+      cacheHit: tiers.cacheHit?.peak?.[k],
+      cacheMiss: tiers.cacheMiss?.peak?.[k],
+      output: tiers.output?.peak?.[k],
+    }
+    if (off.cacheHit === undefined || off.cacheMiss === undefined || off.output === undefined) continue
+    models[id] = {
+      cacheHit: off.cacheHit,
+      cacheMiss: off.cacheMiss,
+      output: off.output,
+      offPeak: off,
+      peak: {
+        cacheHit: pk.cacheHit ?? off.cacheHit,
+        cacheMiss: pk.cacheMiss ?? off.cacheMiss,
+        output: pk.output ?? off.output,
+      },
+    }
+    // 峰谷时代前的历史基础价(官方页面已不再列出,按历史公告数字附带)。
+    const legacy = LEGACY_BASE_PRICES[id]
+    if (legacy !== undefined) models[id].legacyBase = { ...legacy }
+  }
+
+  if (Object.keys(models).length === 0) {
+    // code 供上层按语言渲染提示(见 index.js 的 ERR_NO_MODELS 分支)。
+    const error = new Error('官方页面中未解析出任何模型价格,页面结构可能已变化,请稍后重试或手动编辑价格')
+    error.code = 'ERR_NO_MODELS'
+    throw error
+  }
+  // 生效时间:页面已不再给出(两档方案即时生效)→ null。
+  const effectiveAt = null
+  // 峰时段窗口。
+  let peakWindows = null
+  const plain = stripTags(html)
+  const win = /Peak hours are\s+(.+?)\s+UTC/.exec(plain)
+  if (win !== null) {
+    const pairs = win[1].match(/\d{1,2}:\d{2}/g) ?? []
+    peakWindows = []
+    for (let i = 0; i + 1 < pairs.length; i += 2) {
+      const start = Number(pairs[i].split(':')[0])
+      const end = Number(pairs[i + 1].split(':')[0])
+      if (Number.isFinite(start) && Number.isFinite(end)) peakWindows.push({ start, end })
+    }
+  }
+  return { models, effectiveAt, peakWindows }
+}