From 5ee28cef986e30f19009f0cc381743d674558607 Mon Sep 17 00:00:00 2001 From: Codex Bot Date: Fri, 30 Oct 2026 05:58:54 +0800 Subject: [PATCH] =?UTF-8?q?=E8=A1=A5=E5=85=85Gradium=E7=AC=AC=E4=B8=89?= =?UTF-8?q?=E6=96=B9=E8=AF=AD=E9=9F=B3=E8=AF=84=E6=B5=8B=E5=8F=8A=E6=B5=8B?= =?UTF-8?q?=E8=AF=95=E8=8C=83=E5=9B=B4?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- README.md | 6 +- apps/docs/scripts/auth-cli.ts | 4 +- .../docs/src/components/AICompanyResearch.tsx | 4 +- apps/docs/src/lib/ai-research.ts | 2 +- apps/docs/tests/research.test.ts | 11 ++- content/research/ai-companies.seed.json | 96 ++++++++++++++++++- content/research/ai-income-analysis.json | 2 +- 7 files changed, 109 insertions(+), 16 deletions(-) diff --git a/README.md b/README.md index 5d86536..ad967c7 100644 --- a/README.md +++ b/README.md @@ -98,7 +98,7 @@ AUTH_CONFIG_FILE=/private/tmp/tabularium-auth/config.json AUTH_COOKIE_SECURE=fal 同一版本内容变化会拒绝导入，避免静默读取旧数据。回退代码时继续读取对应旧版本。 所有修改都经过工作分支和 PR，部署仍等待人工合并主线。 -当前资料包含 1,013 条公司/产品记录、10 类业务及待分类组、1,078 条证据记录和 145 条经营观测。 +当前资料包含 1,013 条公司/产品记录、10 类业务及待分类组、1,080 条证据记录和 148 条经营与技术观测。 其中保留 69 家个案资料、928 条目录基础资料，并新增 16 个平台产品/业务样本；目录项不等于完成财务和技术核验。 补充ThinkWheel与中国NaiveAI；MiroMind成立年未核，公开研发年份单列。保留灵初智能成立年排除、深度原理日期待核和Naive近名实体复核记录。 新增7条Speedrun个案，保留标题/正文冲突、七位数数量级、币种不明、合同管线与履历归属排除记录；相对日期不充当同期收入时点。核查页按证据归属链接公司明细，明细反向列出引用其证据的核查；币种不明的声明保留原值，不绘制美元收入长度条。 @@ -137,8 +137,8 @@ MRR×12仅作为显示单位转换，不改写为公司ARR；公司ARR、估算 其中121条原待分类记录已按YC官方业务介绍人工复核主营业务，分类理由保存在公司字段与对应证据摘要中； 仍有3条因简介不足保留待分类。人工复核分类不代表独立验证技术水平、客户效果或财务数据。 AI 关键词可能出现在业务背景中，收入、买家和交付状态仍须逐家深查，不能把目录项的未知收入当作未披露。 -技术专项评审用独立字段记录层级、日期、摘要与证据：未评审、已读技术自述、公开代码/模型材料可访问、有独立测试记录。 -当前仅5条完成该字段的专项记录，其中2条确认材料可访问；没有独立测试记录。缺字段不表示无技术能力，代码或模型卡可读不证明性能。 +技术专项评审用独立字段记录层级、日期、摘要与证据：未评审、已读技术自述、公开代码/模型材料可访问、有第三方独立评测记录。 +当前6条完成该字段的专项记录，其中2条确认材料可访问、1条记录第三方已发表评测（Gradium TTS Sep 2026）；本轮未自行运行模型测试。缺字段不表示无技术能力，代码或模型卡可读不证明性能。 界面支持按资料深度、记录团队规模和技术验证证据筛选；证据列表每页 30 条，避免一次渲染全部来源。 `pnpm test:research` 在远程 Runner 上验证同库持久化、重复读取、版本导入与回退、 diff --git a/apps/docs/scripts/auth-cli.ts b/apps/docs/scripts/auth-cli.ts index 0e68c31..ee20608 100644 --- a/apps/docs/scripts/auth-cli.ts +++ b/apps/docs/scripts/auth-cli.ts @@ -187,8 +187,8 @@ async function main() { const research = await researchResponse.json(); assert.equal(research.companies.length, 1013); assert.equal(research.categories.length, 11); - assert.equal(research.evidence.length, 1078); - assert.equal(research.observations.length, 145); + assert.equal(research.evidence.length, 1080); + assert.equal(research.observations.length, 148); assert.equal((await request("/api/research/ai-companies", { method: "POST", headers })).status, 405); const researchPage = await request("/docs/other-research/ai-company-research/overview", { headers }); assert.equal(researchPage.status, 200); diff --git a/apps/docs/src/components/AICompanyResearch.tsx b/apps/docs/src/components/AICompanyResearch.tsx index 46171c1..4ce641c 100644 --- a/apps/docs/src/components/AICompanyResearch.tsx +++ b/apps/docs/src/components/AICompanyResearch.tsx @@ -11,7 +11,7 @@ const revenueLabels = { self: "公司披露，待核口径", estimate: "仅第 const useLabels: Record = { platform: "支付平台记录", signal: "公司披露信号", estimate: "第三方估算", historical: "历史里程碑", "historical-estimate": "其他时点估算", excluded: "排除收入比较", scale: "团队规模", adoption: "采用规模", - funding: "融资", crosscheck: "同一里程碑的其他资料", + funding: "融资", technical: "第三方技术评测", crosscheck: "同一里程碑的其他资料", }; const money = (value: number, unit = "USD") => unit === "USD" ? new Intl.NumberFormat("zh-CN", { style: "currency", currency: "USD", maximumFractionDigits: 0 }).format(value) @@ -166,7 +166,7 @@ export default function AICompanyResearch() {

{selectedCompany.name}

资料入口　{selectedCompany.extraUrl && 额外核验入口}

主营业务
{selectedCompany.business}
样本组别
{selectedCompany.cohort}
买家
{selectedCompany.buyer}
收费/交付模式
{selectedCompany.businessModel ?? "待进一步核验"}
收费方式
{selectedCompany.price}
交付证据
{selectedCompany.delivery}
团队口径
{selectedCompany.teamBasis ?? "本轮未核验人数"}
分类依据
{selectedCompany.classificationBasis ?? "原有个案分类；随研究复核"}
技术验证证据
{technicalReviewLabels[selectedCompany.technicalReview?.level ?? "unreviewed"]}

{selectedCompany.technicalReview?.summary ?? "本轮未形成专项技术评审记录；不表示公司没有技术能力。"}

{selectedCompany.technicalReview && 评审日期：{selectedCompany.technicalReview.reviewedAt}}{selectedCompany.technicalReview?.testScope &&

测试范围：{selectedCompany.technicalReview.testScope}

}

材料可访问仅确认可查看，不证明性能、可靠性或商业交付质量；不同任务不作统一能力排名。

观测主体
{selectedCompany.entityBasis ?? "公司或经营品牌，法律主体未逐一核验"}
身份与限制
{selectedCompany.notes}
-

经营指标与排除记录

+

经营与技术观测

{!!selectedReviews.length &&
引用该公司证据的核查记录（{selectedReviews.length}）
    {selectedReviews.map((r) =>
  • {r.type}：{r.summary}

    {r.handling}

    {r.evidenceIds.map((id) => { const item = data.evidence.find((e) => e.id === id); return item ? {id} : null; })}
    • )}
} {selectedObservations.length ?
{selectedObservations.map((o) => )}
指标 / 归属原始数值数据时点用途及限制证据
{o.kind}{o.scope}{o.bound !== "=" && o.bound}{money(o.value,o.unit)}{o.timing}{useLabels[o.selection] ?? o.selection}{o.reason}{o.evidence}
:

本轮尚未找到可记录的量化经营指标。收入保持未知。

}

原始证据

diff --git a/apps/docs/src/lib/ai-research.ts b/apps/docs/src/lib/ai-research.ts index 465eeb5..1f48de7 100644 --- a/apps/docs/src/lib/ai-research.ts +++ b/apps/docs/src/lib/ai-research.ts @@ -3,7 +3,7 @@ export const technicalReviewLabels = { unreviewed: "技术证据尚未专项评审", description: "已读技术自述", "public-artifact": "公开代码/模型材料可访问", - "independent-test": "有独立测试记录", + "independent-test": "有第三方独立评测记录", }; export type ResearchCompany = { id: string; name: string; url: string; cat: string; year: number | null; diff --git a/apps/docs/tests/research.test.ts b/apps/docs/tests/research.test.ts index a16cdee..33cc9a6 100644 --- a/apps/docs/tests/research.test.ts +++ b/apps/docs/tests/research.test.ts @@ -24,8 +24,8 @@ test("research revisions share the authentication database and preserve replay l assert.equal(consumeCode(config, "invalid"), "limited"); const initial = readResearchSnapshot(); assert.equal(initial.companies.length, 1013); - assert.equal(initial.evidence.length, 1078); - assert.equal(initial.observations.length, 145); + assert.equal(initial.evidence.length, 1080); + assert.equal(initial.observations.length, 148); assert.deepEqual(readResearchSnapshot(), initial, "reopening does not duplicate records"); const db = openSiteDatabase(); const tables = db.prepare("SELECT name FROM sqlite_master WHERE type='table'").all() as { name: string }[]; @@ -75,9 +75,12 @@ test("disclosure counts exclude client income, biographies, targets and duplicat test("technical reviews keep accessible materials separate from tested capabilities", () => { const snapshot = structuredClone(seed) as ResearchSnapshot; - assert.equal(snapshot.companies.filter((c) => c.technicalReview).length, 5); + assert.equal(snapshot.companies.filter((c) => c.technicalReview).length, 6); assert.equal(snapshot.companies.filter((c) => c.technicalReview?.level === "public-artifact").length, 2); - assert.equal(snapshot.companies.filter((c) => c.technicalReview?.level === "independent-test").length, 0); + assert.equal(snapshot.companies.filter((c) => c.technicalReview?.level === "independent-test").length, 1); + assert.equal(snapshot.companies.find((c) => c.id === "AI051")?.revenueStatus, "unknown"); + assert.match(snapshot.companies.find((c) => c.id === "AI051")!.technicalReview!.testScope!, /Sep 2026/); + assert.ok(snapshot.observations.filter((o) => o.id.startsWith("TECH-GRAD-O-")).every((o) => o.selection === "technical" && o.comparisonGroup === null)); const company = snapshot.companies.find((c) => c.id === "AI061")!; const review = company.technicalReview!; review.evidenceIds = ["missing-technical-evidence"]; diff --git a/content/research/ai-companies.seed.json b/content/research/ai-companies.seed.json index 58472d9..140dc37 100644 --- a/content/research/ai-companies.seed.json +++ b/content/research/ai-companies.seed.json @@ -1,5 +1,5 @@ { - "version": "2026-10-30-speedrun-1", + "version": "2026-10-30-benchmark-1", "asOf": "2026-10-30", "scope": "全球公开资料研究：保留首批个案，并扩展核查2025—2026八个YC批次候选池；成立年份以官方公司页面字段为依据。", "limitations": [ @@ -1317,9 +1317,21 @@ "sourceType": "公司或投资机构资料", "pricingEvidence": "E132", "evidenceIds": [ - "E051" + "E051", + "TECH-GRAD-TTS", + "TECH-GRAD-METHOD" ], - "revenueStatus": "unknown" + "revenueStatus": "unknown", + "technicalReview": { + "level": "independent-test", + "reviewedAt": "2026-10-30", + "summary": "Artificial Analysis公开提供Gradium TTS Sep 2026的Provider Voice Arena偏好评测，Elo 1175，95%区间1162—1188。属于第三方已发表结果，本轮未自行复测。", + "evidenceIds": [ + "TECH-GRAD-TTS", + "TECH-GRAD-METHOD" + ], + "testScope": "评测方Artificial Analysis；模型Gradium TTS (Sep 2026)；英语服务商原生音色、同文本成对听感偏好投票，隐藏模型名。该榜不是固定音色控制实验，也不测端到端语音Agent可靠性、全部语言或企业整体能力。" + } }, { "id": "AI052", @@ -30887,6 +30899,51 @@ "reason": "两周里程碑；日期、币种和经常性定义未核", "bound": "=", "comparisonGroup": null + }, + { + "id": "TECH-GRAD-O-elo", + "company": "AI051", + "name": "Gradium", + "kind": "TTS偏好Elo（95%区间1162—1188）", + "value": 1175, + "unit": "Elo", + "timing": "2026-10-30榜单访问快照；模型标注Sep 2026", + "evidence": "TECH-GRAD-TTS", + "scope": "Gradium TTS (Sep 2026)，Provider Voice Arena", + "selection": "technical", + "reason": "仅同一Provider Voice Arena内有相对意义，非准确率或公司评分", + "bound": "=", + "comparisonGroup": null + }, + { + "id": "TECH-GRAD-O-samples", + "company": "AI051", + "name": "Gradium", + "kind": "页面Samples计数", + "value": 2534, + "unit": "比较样本", + "timing": "2026-10-30榜单访问快照；模型标注Sep 2026", + "evidence": "TECH-GRAD-TTS", + "scope": "Gradium TTS (Sep 2026)，Provider Voice Arena", + "selection": "technical", + "reason": "不是独立付费用户或去重评审人数，不作为客户规模", + "bound": "=", + "comparisonGroup": null + }, + { + "id": "TECH-GRAD-O-price", + "company": "AI051", + "name": "Gradium", + "kind": "第三方折算TTS价格", + "value": 47.8, + "unit": "USD/百万字符", + "timing": "2026-10-30榜单访问快照；模型标注Sep 2026", + "evidence": "TECH-GRAD-TTS", + "scope": "Gradium TTS (Sep 2026)，Provider Voice Arena", + "selection": "technical", + "reason": "按评测方方法折算；不是公司收入、推理成本或毛利", + "bound": "=", + "comparisonGroup": null } ], "evidence": [ @@ -42747,6 +42804,28 @@ "kind": "投资机构资料中的公司自述", "summary": "成立年2026、员工2人；自述两周ARR30.2万（原文$），节时比例未经独立测试。", "limitation": "投资方有利益关系，非独立财务核验；精确收入截止日、币种与经常性定义不完整；管线、客户成果及创始人前公司经历不得作为本公司收入。" + }, + { + "id": "TECH-GRAD-TTS", + "companyId": "AI051", + "title": "Gradium TTS Sep 2026 第三方语音榜单", + "url": "https://artificialanalysis.ai/text-to-speech/leaderboard/provider-voice", + "publishedAt": null, + "accessedAt": "2026-10-30", + "kind": "第三方公开评测", + "summary": "读取时Gradium TTS (Sep 2026) Elo 1175，95%区间1162—1188；Samples为2534，展示8种音色；折算价格47.8美元/百万字符。", + "limitation": "动态榜单的访问快照，不是模型发布日期的评测结果；Samples不等于独立用户数。偏好分数不是正确率，折算价格不是企业毛利或真实客户ARPA。" + }, + { + "id": "TECH-GRAD-METHOD", + "companyId": "AI051", + "title": "Artificial Analysis语音评测方法", + "url": "https://artificialanalysis.ai/methodology/text-to-speech", + "publishedAt": null, + "accessedAt": "2026-10-30", + "kind": "评测方方法说明", + "summary": "同文本成对音频、模型名隐藏后投票；原生音色榜与固定克隆音色榜分开。评分由成对偏好拟合，公布95%区间，榜单多次更新。", + "limitation": "评测方声明不收取上榜或有利结果报酬；本轮未独立审计评测方。原生音色选择影响结果，英语偏好不能推广为全部语言、实时延迟或企业交付能力。" } ], "reviews": [ @@ -43139,6 +43218,17 @@ "SR-E-infragrid" ], "handling": "建立公司与历史经营信号，但不以访问日充当收入时点；同一公司历史阶段不重复计入横截面。" + }, + { + "id": "TECH-GRAD-R-SCOPE", + "type": "技术评测范围", + "company": "Gradium", + "summary": "第三方语音偏好结果仅覆盖榜单所选版本、音色和任务，不是企业整体能力或实际客户经营证据。", + "evidenceIds": [ + "TECH-GRAD-TTS", + "TECH-GRAD-METHOD" + ], + "handling": "记录榜单快照与区间，明确第三方评测、本轮未复测；不将Elo当百分比、Samples当客户数或折算价格当毛利。" } ], "coverage": { diff --git a/content/research/ai-income-analysis.json b/content/research/ai-income-analysis.json index a34d6e3..1b661a8 100644 --- a/content/research/ai-income-analysis.json +++ b/content/research/ai-income-analysis.json @@ -1,5 +1,5 @@ { - "sourceVersion": "2026-10-30-speedrun-1", + "sourceVersion": "2026-10-30-benchmark-1", "asOf": "2026-10-30", "metric": "platform-mrr-usd", "sampleUnit": "product-or-business",