fix(stats): clarify market share providers (#43647)

2026-08-20 10:46:24 -05:00
parent 94e7627de0
commit f27fd08487
4 changed files with 26 additions and 59 deletions
@@ -4,7 +4,7 @@ import { Resource } from "sst/resource"
 import { DatabaseError } from "../database"
 import type { GeoStatMetric } from "./geo"
 import { ModelStatRepo, type ModelStatMetric } from "./model"
-import type { ProviderStatMetric } from "./provider"
+import { statProvider } from "./model-normalization"
 import { DATA_SITE_TIERS, normalizeTier } from "./stat"

 export type UsageProduct = "All Users" | "Zen" | "Go" | "Enterprise"
@@ -140,10 +140,6 @@ type StatMetricRow = Omit<ModelStatMetric, "updatedAt"> & {
  periodStart: number
  updatedAt: number
 }
-type ProviderMetricRow = Omit<ProviderStatMetric, "updatedAt"> & {
-  periodStart: number
-  updatedAt: number
-}
 type GeoMetricRow = Omit<GeoStatMetric, "updatedAt"> & {
  periodStart: number
  updatedAt: number
@@ -171,12 +167,8 @@ type RawRow = Record<string, unknown>
 export function getStatsHomeData(): Effect.Effect<StatsHomeData, StatsDataError> {
  return Effect.tryPromise({
    try: async () => {
-      const [modelRows, providerRows, geoRows] = await Promise.all([
-        listModelDaily(),
-        listProviderDaily(),
-        listGeoDaily(),
-      ])
-      return buildStatsHomeData(modelRows, providerRows, geoRows)
+      const [modelRows, geoRows] = await Promise.all([listModelDaily(), listGeoDaily()])
+      return buildStatsHomeData(modelRows, geoRows)
    },
    catch: (cause) => new StatsDataError(cause),
  })
@@ -241,23 +233,6 @@ async function listModelDaily(): Promise<ModelStatMetric[]> {
  }))
 }

-async function listProviderDaily(): Promise<ProviderStatMetric[]> {
-  return (
-    await queryRows(
-      `select period_key, updated_at, tier, provider, total_tokens from provider_stat
-    where grain = 'day' and client = 'all' and source = 'all'
-    and tier in (${SITE_TIER_PLACEHOLDERS}) order by period_key`,
-      DATA_SITE_TIERS,
-    )
-  ).map((row) => ({
-    periodKey: stringValue(row.period_key),
-    updatedAt: dateValue(row.updated_at),
-    tier: stringValue(row.tier),
-    provider: stringValue(row.provider),
-    totalTokens: numberValue(row.total_tokens),
-  }))
-}
-
 async function listGeoDaily(opts?: { provider?: string; model?: string }): Promise<GeoStatMetric[]> {
  const scope =
    opts?.model && opts.provider
@@ -334,15 +309,10 @@ export const getStatsModelComparisonData = (
    { provider: secondProvider, model: secondModel },
  ])

-function buildStatsHomeData(
-  modelRows: ModelStatMetric[],
-  providerRows: ProviderStatMetric[],
-  geoRows: GeoStatMetric[],
-): StatsHomeData {
+function buildStatsHomeData(modelRows: ModelStatMetric[], geoRows: GeoStatMetric[]): StatsHomeData {
  const normalized = modelRows.flatMap(normalizeStatRow)
-  const providers = providerRows.flatMap(normalizeProviderRow)
  const geo = geoRows.flatMap(normalizeGeoRow)
-  const periods = [...normalized, ...providers, ...geo]
+  const periods = [...normalized, ...geo]
  if (periods.length === 0) return emptyStatsHomeData()

  const earliest = Math.min(...periods.map((row) => row.periodStart))
@@ -377,7 +347,7 @@ function buildStatsHomeData(
    leaderboard: createUsageProductRecord((product) =>
      createRangeRecord((range) => buildLeaderboard(normalized, product, getWindow("1W", earliest, latest))),
    ),
-    market: createRangeRecord((range) => buildMarketShare(providers, "Go", range, getWindow(range, earliest, latest))),
+    market: createRangeRecord((range) => buildMarketShare(normalized, "Go", range, getWindow(range, earliest, latest))),
    tokenCost: createTokenProductRecord((product) =>
      buildTokenCost(normalized, product, getWindow("1W", earliest, latest)),
    ),
@@ -602,15 +572,20 @@ function buildLeaderboard(rows: StatMetricRow[], product: UsageProduct, rankWind
    }))
 }

-function buildMarketShare(rows: ProviderMetricRow[], product: UsageProduct, range: UsageRange, window: DateWindow) {
+function buildMarketShare(rows: StatMetricRow[], product: UsageProduct, range: UsageRange, window: DateWindow) {
+  const providerOrder = aggregateByProvider(rowsForProduct(rows, product, window.start, window.end))
+    .filter((item) => item.provider !== "unknown")
+    .toSorted((a, b) => b.tokens - a.tokens || a.provider.localeCompare(b.provider))
+    .slice(0, 8)
+    .map((item) => item.provider)
+
  return createBuckets(window, range).flatMap((bucket) => {
-    const total = aggregateByProvider(rowsForProduct(rows, product, bucket.start, bucket.end)).toSorted(
-      (a, b) => b.tokens - a.tokens,
-    )
+    const total = aggregateByProvider(rowsForProduct(rows, product, bucket.start, bucket.end))
    const totalTokens = total.reduce((sum, item) => sum + item.tokens, 0)
    if (totalTokens === 0) return []

-    const authors = total.slice(0, 8)
+    const byProvider = new Map(total.map((item) => [item.provider, item.tokens]))
+    const authors = providerOrder.map((provider) => ({ provider, tokens: byProvider.get(provider) ?? 0 }))
    const knownTokens = authors.reduce((sum, item) => sum + item.tokens, 0)
    const withOther = [...authors, { provider: "Other", tokens: Math.max(totalTokens - knownTokens, 0) }].filter(
      (item) => item.tokens > 0,
@@ -773,7 +748,7 @@ function aggregateByModelName(rows: StatMetricRow[]) {
  )
 }

-function aggregateByProvider(rows: ProviderMetricRow[]) {
+function aggregateByProvider(rows: { provider: string; totalTokens: number }[]) {
  return Object.values(
    rows.reduce<Record<string, { provider: string; tokens: number }>>((result, row) => {
      result[row.provider] = {
@@ -917,27 +892,12 @@ function normalizeStatRow(row: ModelStatMetric): StatMetricRow[] {
      periodStart,
      updatedAt,
      tier: normalizeTier(row.tier),
-      provider: row.provider || "unknown",
+      provider: statProvider(row.model, undefined, row.provider) || "unknown",
      model: row.model || "unknown",
    },
  ]
 }

-function normalizeProviderRow(row: ProviderStatMetric): ProviderMetricRow[] {
-  const periodStart = periodKeyTime(row.periodKey)
-  const updatedAt = dateTime(row.updatedAt)
-  if (!Number.isFinite(periodStart) || !Number.isFinite(updatedAt)) return []
-  return [
-    {
-      ...row,
-      periodStart,
-      updatedAt,
-      tier: normalizeTier(row.tier),
-      provider: row.provider || "unknown",
-    },
-  ]
-}
-
 function normalizeGeoRow(row: GeoStatMetric): GeoMetricRow[] {
  const periodStart = periodKeyTime(row.periodKey)
  const updatedAt = dateTime(row.updatedAt)
@@ -948,7 +908,7 @@ function normalizeGeoRow(row: GeoStatMetric): GeoMetricRow[] {
      periodStart,
      updatedAt,
      tier: normalizeTier(row.tier),
-      provider: row.provider || "all",
+      provider: row.provider === "all" ? "all" : statProvider(row.model, undefined, row.provider) || "unknown",
      model: row.model || "all",
      country: row.country || "ZZ",
      continent: row.continent || "",
@@ -987,6 +947,7 @@ function formatProvider(provider: string) {
    deepseek: "DeepSeek",
    google: "Google",
    minimax: "MiniMax",
+    meta: "Meta",
    moonshot: "Moonshot",
    moonshotai: "Moonshot",
    nvidia: "NVIDIA",
@@ -24,6 +24,7 @@ describe("inference stat normalization", () => {
    expect(modelAuthor("kimi-k2.6")).toBe("moonshot")
    expect(modelAuthor("mimo-v2-omni")).toBe("xiaomi")
    expect(modelAuthor("minimax-m2.7")).toBe("minimax")
+    expect(modelAuthor("muse-spark-1.2-contributor")).toBe("meta")
    expect(modelAuthor("nemotron-3-super-free")).toBe("nvidia")
    expect(modelAuthor("qwen3.7-max")).toBe("qwen")
    expect(modelAuthor("alpha-gpt-next")).toBeUndefined()
@@ -67,6 +68,9 @@ describe("inference stat normalization", () => {
      { provider: "openai" },
    ])
    expect(toProviderAggregate(aggregate("big-pickle", "opencode"))).toMatchObject([{ provider: "unknown" }])
+    expect(toProviderAggregate(aggregate("muse-spark-1.2-contributor", "unknown"))).toMatchObject([
+      { provider: "meta" },
+    ])
  })

  test("geo aggregates never keep opencode or big-pickle dimensions", () => {
@@ -9,6 +9,7 @@ export const MODEL_AUTHOR_RULES = [
  { match: "kimi", author: "moonshot" },
  { match: "mimo", author: "xiaomi" },
  { match: "minimax", author: "minimax" },
+  { match: "muse-spark", author: "meta" },
  { match: "nemotron", author: "nvidia" },
  { match: "qwen", author: "qwen" },
 ] as const