diff --git a/apps/server/src/environment/ServerEnvironment.ts b/apps/server/src/environment/ServerEnvironment.ts index 1760623071ef..29e2825b1568 100644 --- a/apps/server/src/environment/ServerEnvironment.ts +++ b/apps/server/src/environment/ServerEnvironment.ts @@ -256,6 +256,7 @@ export const make = Effect.gen(function* () { modelCatalogRefresh: true, usagePriceOverrides: true, usageModelAliases: true, + usageThreadFilter: true, threadPinning: true, threadPinReorder: true, threadActiveReorder: true, diff --git a/apps/server/src/server.ts b/apps/server/src/server.ts index 29e3143cbc63..d7bb814104e6 100644 --- a/apps/server/src/server.ts +++ b/apps/server/src/server.ts @@ -316,6 +316,7 @@ const layerUsage = UsageService.layer.pipe( CursorUsageAccounts.layer.pipe(Layer.provide(CursorAccountReader.layer)), ).pipe(Layer.provide(ProviderHostLive.layer.pipe(Layer.provide(ServerSecretStore.layer)))), ), + Layer.provide(ProjectionStoreV2.layer), Layer.provide(layerServerSettings), ); diff --git a/apps/server/src/usage/UsageService.test.ts b/apps/server/src/usage/UsageService.test.ts index 9a7f7ab82c8f..cf564f9fd248 100644 --- a/apps/server/src/usage/UsageService.test.ts +++ b/apps/server/src/usage/UsageService.test.ts @@ -18,6 +18,7 @@ import { EnvironmentId, ProviderDriverKind, ProviderInstanceId, + ThreadId, UsageDay, type UsageSource, type UsageSummary, @@ -48,6 +49,7 @@ import * as CursorAccountReader from "@t3tools/provider-cursor/server/CursorAcco import * as CursorUsageAccounts from "@t3tools/provider-cursor/server/CursorUsageAccounts"; import * as ServerSecretStore from "../auth/ServerSecretStore.ts"; import * as BackgroundPolicy from "../background/BackgroundPolicy.ts"; +import * as ProjectionStore from "../orchestration-v2/ProjectionStore.ts"; import * as AntigravityUsage from "../provider/Drivers/AntigravityUsage.ts"; import * as ProviderHostLive from "../provider/ProviderHostLive.ts"; import { totalTokens, type UsageRecord } from "@t3tools/provider-core/server/usage"; @@ -196,6 +198,7 @@ const layerService = (input: { readonly platform?: NodeJS.Platform; readonly hostname?: string; readonly secretStore?: Partial; + readonly projectionStore?: Partial; }) => layerCursorUsageAccounts(CursorAccountReader.layer).pipe( Layer.provideMerge(AntigravityUsage.layer), @@ -203,6 +206,7 @@ const layerService = (input: { Layer.provideMerge(ProviderHostLive.layer), Layer.provideMerge(CodexUsageHistory.layer.pipe(Layer.provideMerge(Sqlite.layerMemory))), Layer.provideMerge(Layer.mock(BackgroundPolicy.BackgroundPolicy)({})), + Layer.provideMerge(Layer.mock(ProjectionStore.ProjectionStoreV2)({ ...input.projectionStore })), Layer.provideMerge( Layer.mock(ServerSecretStore.ServerSecretStore)({ get: () => Effect.succeedNone, @@ -3000,6 +3004,95 @@ describe("UsageService", () => { }), ); + it.effect("narrows a summary to a thread's sessions from when it was created", () => + Effect.gen(function* () { + const { transcript, settings, home } = yield* setup; + // A fork's transcript repeats the source's history under the fork's id, then adds its own turn. + yield* Effect.promise(() => + NodeFSP.writeFile( + transcript, + claudeLine(1, 3, "orbit-1", "source") + + claudeLine(2, 5, "orbit-1", "fork") + + claudeLine(3, 7, "orbit-1", "fork").replace("T10:00:00Z", "T12:00:00Z"), + ), + ); + const claudeThread = (nativeId: string | null) => ({ + nativeThreadRef: + nativeId === null + ? null + : { + driver: ProviderDriverKind.make("claudeAgent"), + nativeId, + strength: "strong" as const, + }, + }); + const threads: Record = { + source: { createdAt: "2026-08-01T09:00:00Z", providerThreads: [claudeThread("source")] }, + // Forked at 11:00, with no conversation yet on its second provider. + fork: { + createdAt: "2026-08-01T11:00:00Z", + providerThreads: [claudeThread("fork"), claudeThread(null)], + }, + unstarted: { createdAt: "2026-08-01T09:00:00Z", providerThreads: [] }, + }; + const getThreadRecords = ((threadId: ThreadId) => { + const thread = threads[threadId]; + return thread === undefined + ? Effect.fail(new ProjectionStore.ProjectionStoreThreadNotFoundError({ threadId })) + : Effect.succeed({ + thread: { createdAt: DateTime.makeUnsafe(thread.createdAt) }, + providerThreads: thread.providerThreads, + }); + }) as ProjectionStore.ProjectionStoreV2["Service"]["getThreadRecords"]; + const outputTokens = (summary: UsageSummary) => + summary.buckets.reduce((sum, bucket) => sum + bucket.totals.outputTokens, 0); + + yield* Effect.gen(function* () { + const service = yield* UsageService.make; + const ofThread = (id: string) => + service.readSummary({ ...WINDOW, threadId: ThreadId.make(id) }); + // Same service, concurrent reads: the in-flight key must keep these scans apart. + const [all, source, fork, unstarted, missing] = yield* Effect.all( + [ + service.readSummary(WINDOW), + ofThread("source"), + ofThread("fork"), + // A thread with no conversation yet has no usage, not everyone's. + ofThread("unstarted"), + ofThread("deleted").pipe(Effect.flip), + ], + { concurrency: "unbounded" }, + ); + + assert.strictEqual(outputTokens(all), 15); + assert.strictEqual(outputTokens(source), 3); + assert.strictEqual(outputTokens(fork), 7); + assert.strictEqual(fork.sources[0]?.distinctSessions, 1); + assert.strictEqual(outputTokens(unstarted), 0); + assert.strictEqual(missing.reason, "scanFailed"); + + // A first load asks for the last scan's answer, which is narrowed the same way. + const saved = yield* service.readSummary({ + ...WINDOW, + provisional: true, + threadId: ThreadId.make("fork"), + }); + assert.strictEqual(outputTokens(saved), 7); + assert.isTrue(saved.sources[0]?.refreshing); + }).pipe( + Effect.scoped, + Effect.provide( + layerService({ + prefix: "usage-service-thread-filter-test", + home, + settings, + projectionStore: { getThreadRecords }, + }), + ), + ); + }).pipe(Effect.scoped), + ); + it.effect("reprices unchanged transcripts when custom prices are added, edited, or removed", () => Effect.gen(function* () { const { transcript, settings, home } = yield* setup; diff --git a/apps/server/src/usage/UsageService.ts b/apps/server/src/usage/UsageService.ts index 30edc903a580..5b73ed602032 100644 --- a/apps/server/src/usage/UsageService.ts +++ b/apps/server/src/usage/UsageService.ts @@ -1,4 +1,6 @@ import { + type OrchestrationV2ProviderThread, + type OrchestrationV2ThreadProjection, type ProviderInstanceConfig, ProviderInstanceId, USAGE_CONTRACT_VERSION, @@ -30,6 +32,7 @@ import { HttpClient, HttpClientResponse } from "effect/http"; import { writeFileStringAtomically } from "@t3tools/shared/atomicWrite"; import { formatMalformedUsageWarning } from "@t3tools/shared/usageFormat"; import * as ServerConfig from "../config.ts"; +import * as ProjectionStore from "../orchestration-v2/ProjectionStore.ts"; import * as ServerSettings from "../serverSettings.ts"; import { BUILT_IN_USAGE_DRIVERS, type BuiltInUsageReadersEnv } from "../provider/builtInDrivers.ts"; import type { ProviderDriver } from "@t3tools/provider-core/server/driver"; @@ -39,7 +42,12 @@ import type { TranscriptUsageFormat, UsageRecord, } from "@t3tools/provider-core/server/usage"; -import { canonicalModelFilter, resolveModelAliases, UsageAggregator } from "./usageAggregation.ts"; +import { + canonicalModelFilter, + resolveModelAliases, + UsageAggregator, + type UsageSessions, +} from "./usageAggregation.ts"; import { createOverrideRateTable, isUnlistedModel, @@ -156,6 +164,22 @@ const decodeCachedSources = Schema.decodeUnknownOption( }), ); +/** A summary request once its thread, if any, is resolved to sessions. */ +type UsageScanInput = Omit; + +/** Where a thread's usage is recorded: the native conversations of its provider threads. */ +function threadUsageSessions(records: { + readonly thread: Pick; + readonly providerThreads: ReadonlyArray>; +}): UsageSessions { + return { + ids: new Set( + records.providerThreads.flatMap((thread) => thread.nativeThreadRef?.nativeId ?? []), + ), + sinceTimeMs: DateTime.toEpochMillis(records.thread.createdAt), + }; +} + export class UsageService extends Context.Service< UsageService, { @@ -190,6 +214,7 @@ export const make = Effect.gen(function* () { return id; }); const hostname = yield* HostProcess.Hostname; + const projectionStore = yield* ProjectionStore.ProjectionStoreV2; const readerContext = yield* Effect.context(); const fileCache: ScanCache = new Map(); @@ -823,9 +848,10 @@ export const make = Effect.gen(function* () { }); const scanSummary = Effect.fn("UsageService.scanSummary")(function* ( - input: UsageSummaryInput, + input: UsageScanInput, settings: ServerSettingsValue, provisional: boolean, + sessions: UsageSessions | undefined, ) { if (input.sinceDay > input.untilDay) { return yield* new UsageReadError({ @@ -1007,6 +1033,7 @@ export const make = Effect.gen(function* () { modelAliases, rateModel: (provider, model) => providerRateModels.get(provider)?.(model), ...(input.modelFilter === undefined ? {} : { modelFilter: input.modelFilter }), + ...(sessions === undefined ? {} : { sessions }), }); const windowEndMs = @@ -1187,23 +1214,25 @@ export const make = Effect.gen(function* () { }); const readWindow = Effect.fn("UsageService.readWindow")(function* ( - input: UsageSummaryInput, + input: UsageScanInput, settings: ServerSettingsValue, provisional: boolean, + sessions: UsageSessions | undefined, ) { - const { summary, fromMemory } = yield* scanSummary(input, settings, provisional); + const { summary, fromMemory } = yield* scanSummary(input, settings, provisional, sessions); // With nothing marked refreshing the client would keep a saved answer as final. return fromMemory && !summary.sources.some((source) => source.refreshing === true) - ? (yield* scanSummary(input, settings, false)).summary + ? (yield* scanSummary(input, settings, false, sessions)).summary : summary; }); const inflightScans = new Map>(); const scanKey = ( - input: UsageSummaryInput, + input: UsageScanInput, settings: ServerSettingsValue, provisional: boolean, + sessions: UsageSessions | undefined, ): string => JSON.stringify([ input.timeZone, @@ -1216,17 +1245,32 @@ export const make = Effect.gen(function* () { input.awaitRefresh === true, provisional, input.modelFilter ?? null, + sessions === undefined ? null : [[...sessions.ids].sort(), sessions.sinceTimeMs], ]); const readSummary = Effect.fn("UsageService.readSummary")(function* ( requested: UsageSummaryInput, ) { - const { modelFilter: requestedFilter, ...unfiltered } = requested; + const { modelFilter: requestedFilter, threadId, ...unfiltered } = requested; + const sessions = + threadId === undefined + ? undefined + : yield* projectionStore.getThreadRecords(threadId, ["providerThreads"]).pipe( + Effect.map(threadUsageSessions), + Effect.mapError( + (cause) => + new UsageReadError({ + reason: "scanFailed", + detail: "Could not read the thread's provider sessions", + cause, + }), + ), + ); const modelFilter = canonicalModelFilter(requestedFilter); const input = modelFilter === undefined ? unfiltered : { ...unfiltered, modelFilter }; const settings = yield* readSettings; const provisional = input.provisional === true && input.awaitRefresh !== true; - const key = scanKey(input, settings, provisional); + const key = scanKey(input, settings, provisional, sessions); const deferred = yield* Effect.uninterruptible( Effect.gen(function* () { const existing = inflightScans.get(key); @@ -1237,7 +1281,7 @@ export const make = Effect.gen(function* () { inflightScans.set(key, created); // Detach so caller departure cannot cancel shared scans; finished scans warm the cache. // Register writes before releasing waiters, but start writing after delivering their summary. - yield* readWindow(input, settings, provisional).pipe( + yield* readWindow(input, settings, provisional, sessions).pipe( Effect.onExit((exit) => Effect.sync(() => inflightScans.delete(key)).pipe( Effect.andThen(Exit.isSuccess(exit) ? schedulePersist : Effect.void), diff --git a/apps/server/src/usage/usageAggregation.test.ts b/apps/server/src/usage/usageAggregation.test.ts index 58953fbcfafe..69ba969532c6 100644 --- a/apps/server/src/usage/usageAggregation.test.ts +++ b/apps/server/src/usage/usageAggregation.test.ts @@ -88,6 +88,26 @@ describe("UsageAggregator", () => { expect(bucket).not.toHaveProperty("ultrafastCostUsd"); }); + it("keeps only the chosen sessions' records, before deduplicating", () => { + const aggregator = new UsageAggregator({ + timeZone: "UTC", + sinceDay: "2026-08-01", + untilDay: "2026-08-31", + rates, + sessions: { ids: new Set(["session-fork"]), sinceTimeMs: Date.parse("2026-08-08T00:00:00Z") }, + }); + const forkTurn = { sessionId: "session-fork", timestampMs: Date.parse("2026-08-08T12:00:00Z") }; + // Another session's copy comes first and must not claim the key the fork's record shares. + aggregator.add(record({ dedupeKey: "msg_2:", timestampMs: forkTurn.timestampMs }), "a.jsonl"); + // History the fork copied from its source predates the fork. + aggregator.add(record({ dedupeKey: "msg_1:", sessionId: "session-fork" }), "b.jsonl"); + aggregator.add(record({ dedupeKey: "msg_2:", ...forkTurn }), "b.jsonl"); + + expect(aggregator.finish().buckets).toMatchObject([ + { day: "2026-08-08", sourcePath: "b.jsonl", records: 1, sessions: 1 }, + ]); + }); + it("merges partial copies of a key by channel maximum in the first copy's bucket", () => { const zeroTotals = { uncachedInputTokens: 0, diff --git a/apps/server/src/usage/usageAggregation.ts b/apps/server/src/usage/usageAggregation.ts index f4c39753548f..cfc0dd51d8c4 100644 --- a/apps/server/src/usage/usageAggregation.ts +++ b/apps/server/src/usage/usageAggregation.ts @@ -63,6 +63,13 @@ interface MutableBucket { sessions: Set; } +/** Keeps only these sessions' records at or after `sinceTimeMs`. */ +export interface UsageSessions { + readonly ids: ReadonlySet; + /** A forked session repeats its source's earlier records; this instant leaves them out. */ + readonly sinceTimeMs: number; +} + export interface AggregateOptions { readonly timeZone: string; readonly sinceDay: string; @@ -72,6 +79,7 @@ export interface AggregateOptions { readonly modelAliases?: ReadonlyMap; readonly rateModel?: (provider: UsageRecord["provider"], model: string) => string | undefined; readonly modelFilter?: UsageModelFilter; + readonly sessions?: UsageSessions; readonly resolution?: UsageResolution; readonly sinceTimeMs?: number; readonly untilTimeMs?: number; @@ -248,6 +256,13 @@ export class UsageAggregator { add(input: UsageRecord, sourcePath?: string): boolean { if (!keepsUsageRecord(this.#options.modelFilter, input)) return false; + const sessions = this.#options.sessions; + if ( + sessions !== undefined && + (!sessions.ids.has(input.sessionId) || input.timestampMs < sessions.sinceTimeMs) + ) { + return false; + } const record = this.#mapModel(input); if (record.dedupeKey !== null) { const held = this.#seen.get(record.dedupeKey); diff --git a/apps/web/src/components/chat/ThreadDetailsPanel.tsx b/apps/web/src/components/chat/ThreadDetailsPanel.tsx index d7c309074200..f42a91c6d0a4 100644 --- a/apps/web/src/components/chat/ThreadDetailsPanel.tsx +++ b/apps/web/src/components/chat/ThreadDetailsPanel.tsx @@ -28,7 +28,7 @@ import { ThreadAutomationsPanel } from "./ThreadAutomationsPanel"; import { ThreadRelationshipsPanel } from "./ThreadRelationshipsControl"; import { ThreadServersPanel } from "./ThreadServersPanel"; import { - ThreadDetailsContext, + ThreadDetailsStats, ThreadDetailsUsage, type ThreadDetailsUsageInput, } from "./ThreadDetailsUsage"; @@ -178,6 +178,8 @@ export function ThreadDetailsPanel(props: ThreadDetailsPanelProps) { offersGit && props.activeProjectName !== undefined && (visible.gitActions || (visible.changes && props.isGitRepo)), + context: visible.context ? props.usage.contextWindow : null, + cost: visible.cost && !props.draftId, }; return ( <> @@ -288,8 +290,15 @@ export function ThreadDetailsPanel(props: ThreadDetailsPanelProps) { {visible.usage ? : null} - {visible.context && props.usage.contextWindow ? ( - + {show.context || show.cost ? ( + ) : null} {/* Shown only when no section above has content, so an emptied card can be refilled. */} diff --git a/apps/web/src/components/chat/ThreadDetailsUsage.tsx b/apps/web/src/components/chat/ThreadDetailsUsage.tsx index 2997f6faa21b..5ac0f9bb0b89 100644 --- a/apps/web/src/components/chat/ThreadDetailsUsage.tsx +++ b/apps/web/src/components/chat/ThreadDetailsUsage.tsx @@ -1,9 +1,12 @@ import type { + EnvironmentId, ProviderInstanceId, ServerProvider, ServerProviderUsageWindow, + ThreadId, UsageLimitSourceSnapshots, } from "@t3tools/contracts"; +import { formatPercent, formatUsd } from "@t3tools/shared/usageFormat"; import { collectProviderUsageLimits, type LimitPace, @@ -12,15 +15,17 @@ import { } from "@t3tools/shared/usageLimits"; import { formatDuration, limitsNotice } from "@t3tools/shared/usageLimitsCopy"; import { msg, useGT, useLocale, useMessages } from "gt-react"; -import type { ReactNode } from "react"; +import { Fragment, type ReactNode } from "react"; import { usePrimarySettings } from "../../hooks/useSettings"; import { useNowMinute } from "../../hooks/useNowMinute"; import type { ContextWindowSnapshot } from "@t3tools/client-runtime/state/context-window"; import { formatContextWindowTokens } from "../../lib/contextWindow"; import { cn } from "../../lib/utils"; +import { useThreadUsage } from "../../state/usage"; import { formatUpcomingTimestamp } from "../../timestampFormat"; import { Tooltip, TooltipPopup, TooltipTrigger } from "../ui/tooltip"; +import { costTypeSegments } from "../usage/usageBreakdown"; import { formatContextWindowCompactionMessage } from "./compaction.logic"; import { ThreadDetailsSection } from "./ThreadDetailsSection"; @@ -84,8 +89,32 @@ export function ThreadDetailsUsage({ ); } -/** The thread's context window. The bar fills as context is used; the tick marks auto-compaction. */ -export function ThreadDetailsContext({ usage }: { readonly usage: ContextWindowSnapshot }) { +/** The thread's own figures. They share one grid, so their labels and values line up. */ +export function ThreadDetailsStats({ + contextWindow, + cost, +}: { + readonly contextWindow: ContextWindowSnapshot | null; + /** The thread to price. Null hides the cost row. */ + readonly cost: { readonly environmentId: EnvironmentId; readonly threadId: ThreadId } | null; +}) { + const gt = useGT(); + return ( + +
+ {contextWindow ? : null} + {cost ? : null} +
+
+ ); +} + +/** The bar fills as context is used; the tick marks auto-compaction. */ +function ContextRow({ usage }: { readonly usage: ContextWindowSnapshot }) { const maxTokens = usage.maxTokens != null && usage.maxTokens > 0 ? usage.maxTokens : null; const used = maxTokens !== null && usage.usedPercentage !== null @@ -101,56 +130,124 @@ export function ThreadDetailsContext({ usage }: { readonly usage: ContextWindowS const locale = useLocale(); const usedTokens = formatContextWindowTokens(usage.usedTokens); return ( - -
- = 90 ? "error" : used >= 80 ? "warning" : null} - tick={tick} - muted - value={ - maxTokens !== null - ? `${usedTokens} / ${formatContextWindowTokens(maxTokens)}` - : usedTokens - } - ariaValue={ - used !== null - ? gt("{used}% used", { used }) - : gt("{usedTokens} tokens used", { usedTokens }) - } - tooltip={ - <> - - {used !== null - ? gt("Context: {used}% used", { used }) - : gt("Context: {usedTokens} tokens used", { usedTokens })} - - {maxTokens !== null ? ( - - {gt("{used} of {max} tokens", { - used: usage.usedTokens.toLocaleString(locale), - max: maxTokens.toLocaleString(locale), - })} - - ) : null} - {threshold !== null && threshold > 0 ? ( - - {formatContextWindowCompactionMessage(threshold, m)} - - ) : null} - - } - /> -
-
+ = 90 ? "error" : used >= 80 ? "warning" : null} + tick={tick} + muted + value={ + maxTokens !== null ? `${usedTokens} / ${formatContextWindowTokens(maxTokens)}` : usedTokens + } + ariaValue={ + used !== null + ? gt("{used}% used", { used }) + : gt("{usedTokens} tokens used", { usedTokens }) + } + tooltip={ + <> + + {used !== null + ? gt("Context: {used}% used", { used }) + : gt("Context: {usedTokens} tokens used", { usedTokens })} + + {maxTokens !== null ? ( + + {gt("{used} of {max} tokens", { + used: usage.usedTokens.toLocaleString(locale), + max: maxTokens.toLocaleString(locale), + })} + + ) : null} + {threshold !== null && threshold > 0 ? ( + + {formatContextWindowCompactionMessage(threshold, m)} + + ) : null} + + } + /> + ); +} + +/** + * What the thread's model usage would cost at API prices, as the Usage tab prices it. A figure, + * not a meter, so it has no bar. Renders nothing until the server reports usage for the thread. + */ +function CostRow({ + environmentId, + threadId, +}: { + readonly environmentId: EnvironmentId; + readonly threadId: ThreadId; +}) { + const gt = useGT(); + const m = useMessages(); + const locale = useLocale(); + const usage = useThreadUsage({ environmentId, threadId }); + // No records means the provider's usage is not read here, which is not a cost of zero. + if (usage === null || usage.records === 0) return null; + const tokens = gt("{tokens} tokens", { tokens: formatContextWindowTokens(usage.totalTokens) }); + const label = gt("Cost", { $context: "thread cost row label" }); + const { unpricedShare } = usage.costQuality; + if (unpricedShare >= 1) { + return ( + + {gt("Cost: no known price for this model")} + {tokens} + + } + /> + ); + } + const value = formatCost(usage.costUsd, locale); + // A part that rounds to no cents says nothing next to the total. + const parts = costTypeSegments(usage.categoryCost).filter((part) => part.value >= 0.005); + return ( + + {gt("Cost: {value} · API estimate", { value })} + {parts.length > 0 ? ( + + {parts.map((part) => ( + + {m(part.label)} + + {formatUsd(part.value, locale)} + + + ))} + + ) : null} + {tokens} + {unpricedShare > 0 ? ( + + {gt("Excludes {share} unpriced records", { share: formatPercent(unpricedShare) })} + + ) : null} + + } + /> ); } +/** A cost that rounds to no cents still cost something. */ +function formatCost(usd: number, locale: string): string { + return usd > 0 && usd < 0.005 ? `<${formatUsd(0.01, locale)}` : formatUsd(usd, locale); +} + function LimitRow({ window, now, @@ -201,6 +298,7 @@ function UsageRow({ ariaValue = value, tooltip, muted = false, + dimValue, }: { readonly label: string; /** Bar fill from 0 to 100. Null hides the bar. */ @@ -212,6 +310,8 @@ function UsageRow({ readonly ariaValue?: string; readonly tooltip: ReactNode; readonly muted?: boolean; + /** A row without a bar dims its value, as a notice. False keeps it a figure. */ + readonly dimValue?: boolean; }) { const gt = useGT(); return ( @@ -261,7 +361,7 @@ function UsageRow({ ? "text-error-foreground" : tone === "warning" ? "text-warning-foreground" - : fill === null + : (dimValue ?? fill === null) ? "text-muted-foreground" : "text-foreground/80", )} diff --git a/apps/web/src/components/chat/threadDetailsItems.ts b/apps/web/src/components/chat/threadDetailsItems.ts index 9f8f05449c55..c55f64db50b6 100644 --- a/apps/web/src/components/chat/threadDetailsItems.ts +++ b/apps/web/src/components/chat/threadDetailsItems.ts @@ -27,6 +27,11 @@ export const THREAD_DETAILS_ITEMS = [ { id: "sideChats", label: msg("Side chats"), defaultVisible: true }, { id: "usage", label: msg("Usage limits"), defaultVisible: false }, { id: "context", label: msg("Context window"), defaultVisible: false }, + { + id: "cost", + label: msg("Cost", { $context: "thread details item: what the thread cost" }), + defaultVisible: false, + }, ] as const; export type ThreadDetailsItemId = (typeof THREAD_DETAILS_ITEMS)[number]["id"]; diff --git a/apps/web/src/i18n/translations/de.json b/apps/web/src/i18n/translations/de.json index 648fd58859db..24ee14abf337 100644 --- a/apps/web/src/i18n/translations/de.json +++ b/apps/web/src/i18n/translations/de.json @@ -4030,13 +4030,19 @@ "90f977e97ae8b6c6": "Unter dem Tempo", "b560c47847c88b2e": "Limits", "42a02a4b27828027": "Nicht verfügbar", - "16fb3b76d0abbb2e": "Kontextfenster", + "5cc0aae28ea5e189": "Kontext und Kosten", "83bbceecef0f6872": "Kontext", "d996426072afbeb2": "{used} % genutzt", "ce0545a76897e406": "{usedTokens} Tokens genutzt", "7c1d4c58aeb9e1b3": "Kontext: {used} % genutzt", "4214d97447ddd9e9": "Kontext: {usedTokens} Tokens genutzt", "66be4fb43249bbd2": "{used} von {max} Tokens", + "11a5a798041da0a8": "{tokens} Tokens", + "5db9c60b6d8268b8": "Kosten", + "fce17caaa0d1a45b": "Ohne Preis", + "4d68d68b09ccdf8c": "Kosten: kein bekannter Preis für dieses Modell", + "3aa1c0ff0fde4350": "Kosten: {value} · API-Schätzung", + "4071b0b3c24bffe3": "Ohne {share} Einträge ohne Preis", "ee3ae58c8b38a425": "{left} % übrig", "ebb26f93fb83a71b": "{label}: {left} % übrig", "728c6cc1431e76c8": "Wird in {duration} zurückgesetzt · {time}", @@ -4536,6 +4542,8 @@ "c0910cc360603b7c": "Änderungen", "b42b2e806ba09c36": "Server", "ba1b076eac28418c": "Abstammung", + "16fb3b76d0abbb2e": "Kontextfenster", + "063c45f6f2464936": "Kosten", "ad65d733fa96f01a": "Bearbeitet {label}", "329ae56f86cc3cef": "Sucht im Web nach {label}", "c63f0fdfb05dc589": "Sucht nach {label}", @@ -11818,7 +11826,6 @@ "8bd6f07136317c8f": "Kein Anbieter in den ausgewählten Umgebungen meldet Abo-Limits.", "87f5d0c61b335f38": "Nutzung in ChatGPT mit deinem verbundenen Konto ansehen.", "5b4c71b36b979e80": "Kosten", - "fce17caaa0d1a45b": "Ohne Preis", "8f01f72ab87d0628": "Tokens", "bad042fdd1b3cb1f": "Pro 1 Mio. Tokens", "c9903664b61c8e25": "Cache-Treffer", diff --git a/apps/web/src/i18n/translations/en.json b/apps/web/src/i18n/translations/en.json index 599944358b5b..4c89dbe5f619 100644 --- a/apps/web/src/i18n/translations/en.json +++ b/apps/web/src/i18n/translations/en.json @@ -4029,13 +4029,19 @@ "90f977e97ae8b6c6": "Under pace", "b560c47847c88b2e": "Limits", "42a02a4b27828027": "Unavailable", - "16fb3b76d0abbb2e": "Context window", + "5cc0aae28ea5e189": "Context and cost", "83bbceecef0f6872": "Context", "d996426072afbeb2": "{used}% used", "ce0545a76897e406": "{usedTokens} tokens used", "7c1d4c58aeb9e1b3": "Context: {used}% used", "4214d97447ddd9e9": "Context: {usedTokens} tokens used", "66be4fb43249bbd2": "{used} of {max} tokens", + "11a5a798041da0a8": "{tokens} tokens", + "5db9c60b6d8268b8": "Cost", + "fce17caaa0d1a45b": "Unpriced", + "4d68d68b09ccdf8c": "Cost: no known price for this model", + "3aa1c0ff0fde4350": "Cost: {value} · API estimate", + "4071b0b3c24bffe3": "Excludes {share} unpriced records", "ee3ae58c8b38a425": "{left}% left", "ebb26f93fb83a71b": "{label}: {left}% left", "728c6cc1431e76c8": "Resets in {duration} · {time}", @@ -4535,6 +4541,8 @@ "c0910cc360603b7c": "Changes", "b42b2e806ba09c36": "Servers", "ba1b076eac28418c": "Lineage", + "16fb3b76d0abbb2e": "Context window", + "063c45f6f2464936": "Cost", "ad65d733fa96f01a": "Editing {label}", "329ae56f86cc3cef": "Searching the web for {label}", "c63f0fdfb05dc589": "Searching for {label}", @@ -11826,7 +11834,6 @@ "8bd6f07136317c8f": "No provider on the selected environments reports subscription limits.", "87f5d0c61b335f38": "View usage in ChatGPT with your connected account.", "5b4c71b36b979e80": "Cost", - "fce17caaa0d1a45b": "Unpriced", "8f01f72ab87d0628": "Tokens", "bad042fdd1b3cb1f": "Per 1M tokens", "c9903664b61c8e25": "Cache hit", diff --git a/apps/web/src/i18n/translations/es.json b/apps/web/src/i18n/translations/es.json index 956f4e17a961..dea344fb3b10 100644 --- a/apps/web/src/i18n/translations/es.json +++ b/apps/web/src/i18n/translations/es.json @@ -4027,13 +4027,19 @@ "90f977e97ae8b6c6": "Por debajo del ritmo", "b560c47847c88b2e": "Límites", "42a02a4b27828027": "No disponible", - "16fb3b76d0abbb2e": "Ventana de contexto", + "5cc0aae28ea5e189": "Contexto y coste", "83bbceecef0f6872": "Contexto", "d996426072afbeb2": "{used} % usado", "ce0545a76897e406": "{usedTokens} tokens usados", "7c1d4c58aeb9e1b3": "Contexto: {used} % usado", "4214d97447ddd9e9": "Contexto: {usedTokens} tokens usados", "66be4fb43249bbd2": "{used} de {max} tokens", + "11a5a798041da0a8": "{tokens} tokens", + "5db9c60b6d8268b8": "Coste", + "fce17caaa0d1a45b": "Sin precio", + "4d68d68b09ccdf8c": "Coste: no hay precio conocido para este modelo", + "3aa1c0ff0fde4350": "Coste: {value} · Estimación de API", + "4071b0b3c24bffe3": "Excluye un {share} de registros sin precio", "ee3ae58c8b38a425": "{left} % restante", "ebb26f93fb83a71b": "{label}: {left} % restante", "728c6cc1431e76c8": "Se restablece en {duration} · {time}", @@ -4533,6 +4539,8 @@ "c0910cc360603b7c": "Cambios", "b42b2e806ba09c36": "Servidores", "ba1b076eac28418c": "Linaje", + "16fb3b76d0abbb2e": "Ventana de contexto", + "063c45f6f2464936": "Coste", "ad65d733fa96f01a": "Editando {label}", "329ae56f86cc3cef": "Buscando {label} en la web", "c63f0fdfb05dc589": "Buscando {label}", @@ -11811,7 +11819,6 @@ "8bd6f07136317c8f": "Ningún proveedor de los entornos seleccionados informa de límites de suscripción.", "87f5d0c61b335f38": "Consulta el uso en ChatGPT con tu cuenta conectada.", "5b4c71b36b979e80": "Coste", - "fce17caaa0d1a45b": "Sin precio", "8f01f72ab87d0628": "Tokens", "bad042fdd1b3cb1f": "Por 1 M de tokens", "c9903664b61c8e25": "Acierto de caché", diff --git a/apps/web/src/i18n/translations/fr.json b/apps/web/src/i18n/translations/fr.json index 1c982cc6d6f5..d7c515684d87 100644 --- a/apps/web/src/i18n/translations/fr.json +++ b/apps/web/src/i18n/translations/fr.json @@ -4028,13 +4028,19 @@ "90f977e97ae8b6c6": "En dessous du rythme", "b560c47847c88b2e": "Limites", "42a02a4b27828027": "Indisponible", - "16fb3b76d0abbb2e": "Fenêtre de contexte", + "5cc0aae28ea5e189": "Contexte et coût", "83bbceecef0f6872": "Contexte", "d996426072afbeb2": "{used} % utilisé", "ce0545a76897e406": "{usedTokens} tokens utilisés", "7c1d4c58aeb9e1b3": "Contexte : {used} % utilisé", "4214d97447ddd9e9": "Contexte : {usedTokens} tokens utilisés", "66be4fb43249bbd2": "{used} tokens sur {max}", + "11a5a798041da0a8": "{tokens} tokens", + "5db9c60b6d8268b8": "Coût", + "fce17caaa0d1a45b": "Sans prix", + "4d68d68b09ccdf8c": "Coût : aucun prix connu pour ce modèle", + "3aa1c0ff0fde4350": "Coût : {value} · Estimation API", + "4071b0b3c24bffe3": "Hors {share} d’enregistrements sans prix", "ee3ae58c8b38a425": "{left} % restant", "ebb26f93fb83a71b": "{label} : {left} % restant", "728c6cc1431e76c8": "Réinitialisation dans {duration} · {time}", @@ -4534,6 +4540,8 @@ "c0910cc360603b7c": "Modifications", "b42b2e806ba09c36": "Serveurs", "ba1b076eac28418c": "Lignée", + "16fb3b76d0abbb2e": "Fenêtre de contexte", + "063c45f6f2464936": "Coût", "ad65d733fa96f01a": "Modification de {label}", "329ae56f86cc3cef": "Recherche web de {label}", "c63f0fdfb05dc589": "Recherche de {label}", @@ -11816,7 +11824,6 @@ "8bd6f07136317c8f": "Aucun fournisseur des environnements sélectionnés ne signale de limites d’abonnement.", "87f5d0c61b335f38": "Consultez l’utilisation dans ChatGPT avec votre compte connecté.", "5b4c71b36b979e80": "Coût", - "fce17caaa0d1a45b": "Sans prix", "8f01f72ab87d0628": "Tokens", "bad042fdd1b3cb1f": "Pour 1 M de tokens", "c9903664b61c8e25": "Succès de cache", diff --git a/apps/web/src/i18n/translations/ja.json b/apps/web/src/i18n/translations/ja.json index 879531a396d2..043d245a7e68 100644 --- a/apps/web/src/i18n/translations/ja.json +++ b/apps/web/src/i18n/translations/ja.json @@ -4018,13 +4018,19 @@ "90f977e97ae8b6c6": "ペースより遅い", "b560c47847c88b2e": "上限", "42a02a4b27828027": "利用不可", - "16fb3b76d0abbb2e": "コンテキストウィンドウ", + "5cc0aae28ea5e189": "コンテキストとコスト", "83bbceecef0f6872": "コンテキスト", "d996426072afbeb2": "{used}% 使用", "ce0545a76897e406": "{usedTokens} トークン使用", "7c1d4c58aeb9e1b3": "コンテキスト: {used}% 使用", "4214d97447ddd9e9": "コンテキスト: {usedTokens} トークン使用", "66be4fb43249bbd2": "{used} / {max} トークン", + "11a5a798041da0a8": "{tokens} トークン", + "5db9c60b6d8268b8": "コスト", + "fce17caaa0d1a45b": "価格不明", + "4d68d68b09ccdf8c": "コスト: このモデルの価格は不明です", + "3aa1c0ff0fde4350": "コスト: {value} · API の推定額", + "4071b0b3c24bffe3": "価格不明の記録 {share} を除く", "ee3ae58c8b38a425": "残り {left}%", "ebb26f93fb83a71b": "{label}: 残り {left}%", "728c6cc1431e76c8": "{duration} 後にリセット · {time}", @@ -4523,6 +4529,8 @@ "c0910cc360603b7c": "変更", "b42b2e806ba09c36": "サーバー", "ba1b076eac28418c": "系譜", + "16fb3b76d0abbb2e": "コンテキストウィンドウ", + "063c45f6f2464936": "コスト", "ad65d733fa96f01a": "{label} を編集中", "329ae56f86cc3cef": "Web で {label} を検索中", "c63f0fdfb05dc589": "{label} を検索中", @@ -11766,7 +11774,6 @@ "8bd6f07136317c8f": "選択した環境には、サブスクリプションの上限を報告するプロバイダーがありません。", "87f5d0c61b335f38": "接続したアカウントで ChatGPT の使用量を表示します。", "5b4c71b36b979e80": "コスト", - "fce17caaa0d1a45b": "価格不明", "8f01f72ab87d0628": "トークン", "bad042fdd1b3cb1f": "100 万トークンあたり", "c9903664b61c8e25": "キャッシュヒット", diff --git a/apps/web/src/i18n/translations/ko.json b/apps/web/src/i18n/translations/ko.json index fae8ce2ec6fc..a514d9da61cd 100644 --- a/apps/web/src/i18n/translations/ko.json +++ b/apps/web/src/i18n/translations/ko.json @@ -4015,13 +4015,19 @@ "90f977e97ae8b6c6": "예상보다 느림", "b560c47847c88b2e": "한도", "42a02a4b27828027": "사용할 수 없음", - "16fb3b76d0abbb2e": "컨텍스트 창", + "5cc0aae28ea5e189": "컨텍스트 및 비용", "83bbceecef0f6872": "컨텍스트", "d996426072afbeb2": "{used}% 사용됨", "ce0545a76897e406": "토큰 {usedTokens}개 사용됨", "7c1d4c58aeb9e1b3": "컨텍스트: {used}% 사용됨", "4214d97447ddd9e9": "컨텍스트: 토큰 {usedTokens}개 사용됨", "66be4fb43249bbd2": "토큰 {max}개 중 {used}개", + "11a5a798041da0a8": "토큰 {tokens}개", + "5db9c60b6d8268b8": "비용", + "fce17caaa0d1a45b": "가격 없음", + "4d68d68b09ccdf8c": "비용: 이 모델의 알려진 가격이 없습니다", + "3aa1c0ff0fde4350": "비용: {value} · API 추정치", + "4071b0b3c24bffe3": "가격 없는 기록 {share} 제외", "ee3ae58c8b38a425": "{left}% 남음", "ebb26f93fb83a71b": "{label}: {left}% 남음", "728c6cc1431e76c8": "{duration} 후 초기화 · {time}", @@ -4520,6 +4526,8 @@ "c0910cc360603b7c": "변경 사항", "b42b2e806ba09c36": "서버", "ba1b076eac28418c": "계보", + "16fb3b76d0abbb2e": "컨텍스트 창", + "063c45f6f2464936": "비용", "ad65d733fa96f01a": "{label} 편집 중", "329ae56f86cc3cef": "웹에서 {label} 검색 중", "c63f0fdfb05dc589": "{label} 검색 중", @@ -11763,7 +11771,6 @@ "8bd6f07136317c8f": "선택한 환경에 구독 한도를 보고하는 제공자가 없습니다.", "87f5d0c61b335f38": "연결된 계정으로 ChatGPT에서 사용량을 확인하세요.", "5b4c71b36b979e80": "비용", - "fce17caaa0d1a45b": "가격 없음", "8f01f72ab87d0628": "토큰", "bad042fdd1b3cb1f": "100만 토큰당", "c9903664b61c8e25": "캐시 적중", diff --git a/apps/web/src/i18n/translations/pt-BR.json b/apps/web/src/i18n/translations/pt-BR.json index 7010bef288b1..d94018f550ce 100644 --- a/apps/web/src/i18n/translations/pt-BR.json +++ b/apps/web/src/i18n/translations/pt-BR.json @@ -4027,13 +4027,19 @@ "90f977e97ae8b6c6": "Abaixo do ritmo", "b560c47847c88b2e": "Limites", "42a02a4b27828027": "Indisponível", - "16fb3b76d0abbb2e": "Janela de contexto", + "5cc0aae28ea5e189": "Contexto e custo", "83bbceecef0f6872": "Contexto", "d996426072afbeb2": "{used}% usado", "ce0545a76897e406": "{usedTokens} tokens usados", "7c1d4c58aeb9e1b3": "Contexto: {used}% usado", "4214d97447ddd9e9": "Contexto: {usedTokens} tokens usados", "66be4fb43249bbd2": "{used} de {max} tokens", + "11a5a798041da0a8": "{tokens} tokens", + "5db9c60b6d8268b8": "Custo", + "fce17caaa0d1a45b": "Sem preço", + "4d68d68b09ccdf8c": "Custo: não há preço conhecido para este modelo", + "3aa1c0ff0fde4350": "Custo: {value} · Estimativa da API", + "4071b0b3c24bffe3": "Exclui {share} de registros sem preço", "ee3ae58c8b38a425": "{left}% restante", "ebb26f93fb83a71b": "{label}: {left}% restante", "728c6cc1431e76c8": "Renova em {duration} · {time}", @@ -4533,6 +4539,8 @@ "c0910cc360603b7c": "Alterações", "b42b2e806ba09c36": "Servidores", "ba1b076eac28418c": "Linhagem", + "16fb3b76d0abbb2e": "Janela de contexto", + "063c45f6f2464936": "Custo", "ad65d733fa96f01a": "Editando {label}", "329ae56f86cc3cef": "Pesquisando {label} na web", "c63f0fdfb05dc589": "Pesquisando {label}", @@ -11815,7 +11823,6 @@ "8bd6f07136317c8f": "Nenhum provedor nos ambientes selecionados informa limites de assinatura.", "87f5d0c61b335f38": "Veja o uso no ChatGPT com sua conta conectada.", "5b4c71b36b979e80": "Custo", - "fce17caaa0d1a45b": "Sem preço", "8f01f72ab87d0628": "Tokens", "bad042fdd1b3cb1f": "Por 1M de tokens", "c9903664b61c8e25": "Acerto de cache", diff --git a/apps/web/src/i18n/translations/ru.json b/apps/web/src/i18n/translations/ru.json index 0766d953dd02..56eb57562231 100644 --- a/apps/web/src/i18n/translations/ru.json +++ b/apps/web/src/i18n/translations/ru.json @@ -4027,13 +4027,19 @@ "90f977e97ae8b6c6": "Медленнее темпа", "b560c47847c88b2e": "Лимиты", "42a02a4b27828027": "Недоступно", - "16fb3b76d0abbb2e": "Контекстное окно", + "5cc0aae28ea5e189": "Контекст и стоимость", "83bbceecef0f6872": "Контекст", "d996426072afbeb2": "Использовано {used}%", "ce0545a76897e406": "Использовано токенов: {usedTokens}", "7c1d4c58aeb9e1b3": "Контекст: использовано {used}%", "4214d97447ddd9e9": "Контекст: использовано токенов: {usedTokens}", "66be4fb43249bbd2": "{used} из {max} токенов", + "11a5a798041da0a8": "Токенов: {tokens}", + "5db9c60b6d8268b8": "Стоимость", + "fce17caaa0d1a45b": "Без цены", + "4d68d68b09ccdf8c": "Стоимость: для этой модели нет известной цены", + "3aa1c0ff0fde4350": "Стоимость: {value} · Оценка по API", + "4071b0b3c24bffe3": "Не включает {share} записей без цены", "ee3ae58c8b38a425": "Осталось {left}%", "ebb26f93fb83a71b": "{label}: осталось {left}%", "728c6cc1431e76c8": "Сброс через {duration} · {time}", @@ -4533,6 +4539,8 @@ "c0910cc360603b7c": "Изменения", "b42b2e806ba09c36": "Серверы", "ba1b076eac28418c": "Родословная", + "16fb3b76d0abbb2e": "Контекстное окно", + "063c45f6f2464936": "Стоимость", "ad65d733fa96f01a": "Редактирование {label}", "329ae56f86cc3cef": "Поиск в интернете: {label}", "c63f0fdfb05dc589": "Поиск: {label}", @@ -11816,7 +11824,6 @@ "8bd6f07136317c8f": "Ни один провайдер в выбранных окружениях не сообщает лимиты подписки.", "87f5d0c61b335f38": "Смотрите использование в ChatGPT через подключённый аккаунт.", "5b4c71b36b979e80": "Стоимость", - "fce17caaa0d1a45b": "Без цены", "8f01f72ab87d0628": "Токены", "bad042fdd1b3cb1f": "За 1 млн токенов", "c9903664b61c8e25": "Попадания в кеш", diff --git a/apps/web/src/i18n/translations/sv.json b/apps/web/src/i18n/translations/sv.json index 5e6d1607edf3..f9685e41bc85 100644 --- a/apps/web/src/i18n/translations/sv.json +++ b/apps/web/src/i18n/translations/sv.json @@ -4028,13 +4028,19 @@ "90f977e97ae8b6c6": "Under takten", "b560c47847c88b2e": "Gränser", "42a02a4b27828027": "Inte tillgänglig", - "16fb3b76d0abbb2e": "Kontextfönster", + "5cc0aae28ea5e189": "Kontext och kostnad", "83bbceecef0f6872": "Kontext", "d996426072afbeb2": "{used} % använt", "ce0545a76897e406": "{usedTokens} token använda", "7c1d4c58aeb9e1b3": "Kontext: {used} % använt", "4214d97447ddd9e9": "Kontext: {usedTokens} token använda", "66be4fb43249bbd2": "{used} av {max} token", + "11a5a798041da0a8": "{tokens} token", + "5db9c60b6d8268b8": "Kostnad", + "fce17caaa0d1a45b": "Utan pris", + "4d68d68b09ccdf8c": "Kostnad: inget känt pris för den här modellen", + "3aa1c0ff0fde4350": "Kostnad: {value} · API-uppskattning", + "4071b0b3c24bffe3": "Exkluderar {share} poster utan pris", "ee3ae58c8b38a425": "{left} % kvar", "ebb26f93fb83a71b": "{label}: {left} % kvar", "728c6cc1431e76c8": "Återställs om {duration} · {time}", @@ -4534,6 +4540,8 @@ "c0910cc360603b7c": "Ändringar", "b42b2e806ba09c36": "Servrar", "ba1b076eac28418c": "Härkomst", + "16fb3b76d0abbb2e": "Kontextfönster", + "063c45f6f2464936": "Kostnad", "ad65d733fa96f01a": "Redigerar {label}", "329ae56f86cc3cef": "Söker på webben efter {label}", "c63f0fdfb05dc589": "Söker efter {label}", @@ -11824,7 +11832,6 @@ "8bd6f07136317c8f": "Ingen leverantör i de valda miljöerna rapporterar abonnemangsgränser.", "87f5d0c61b335f38": "Visa användning i ChatGPT med ditt anslutna konto.", "5b4c71b36b979e80": "Kostnad", - "fce17caaa0d1a45b": "Utan pris", "8f01f72ab87d0628": "Token", "bad042fdd1b3cb1f": "Per 1 M token", "c9903664b61c8e25": "Cacheträff", diff --git a/apps/web/src/i18n/translations/zh.json b/apps/web/src/i18n/translations/zh.json index c86b9e6c55e4..36e23a2c7432 100644 --- a/apps/web/src/i18n/translations/zh.json +++ b/apps/web/src/i18n/translations/zh.json @@ -4034,13 +4034,19 @@ "90f977e97ae8b6c6": "用量消耗偏慢", "b560c47847c88b2e": "限制", "42a02a4b27828027": "不可用", - "16fb3b76d0abbb2e": "上下文窗口", + "5cc0aae28ea5e189": "上下文和费用", "83bbceecef0f6872": "上下文", "d996426072afbeb2": "已使用 {used}%", "ce0545a76897e406": "已使用 {usedTokens} token", "7c1d4c58aeb9e1b3": "上下文:已使用 {used}%", "4214d97447ddd9e9": "上下文:已使用 {usedTokens} token", "66be4fb43249bbd2": "{used} / {max} token", + "11a5a798041da0a8": "{tokens} token", + "5db9c60b6d8268b8": "费用", + "fce17caaa0d1a45b": "未定价", + "4d68d68b09ccdf8c": "费用:此模型没有已知价格", + "3aa1c0ff0fde4350": "费用:{value} · API 估算", + "4071b0b3c24bffe3": "不含 {share} 的未定价记录", "ee3ae58c8b38a425": "剩余 {left}%", "ebb26f93fb83a71b": "{label}:剩余 {left}%", "728c6cc1431e76c8": "{duration} 后重置 · {time}", @@ -4540,6 +4546,8 @@ "c0910cc360603b7c": "改动", "b42b2e806ba09c36": "服务器", "ba1b076eac28418c": "会话关系", + "16fb3b76d0abbb2e": "上下文窗口", + "063c45f6f2464936": "费用", "ad65d733fa96f01a": "正在编辑 {label}", "329ae56f86cc3cef": "正在网上搜索 {label}", "c63f0fdfb05dc589": "正在搜索 {label}", @@ -11841,7 +11849,6 @@ "8bd6f07136317c8f": "选中的环境上没有提供方报告订阅限制。", "87f5d0c61b335f38": "用已连接的账户在 ChatGPT 中查看用量。", "5b4c71b36b979e80": "费用", - "fce17caaa0d1a45b": "未定价", "8f01f72ab87d0628": "token", "bad042fdd1b3cb1f": "每百万 token", "c9903664b61c8e25": "缓存命中", diff --git a/apps/web/src/state/usage.ts b/apps/web/src/state/usage.ts index 446bd0f4647a..5bb5ce05c004 100644 --- a/apps/web/src/state/usage.ts +++ b/apps/web/src/state/usage.ts @@ -3,13 +3,21 @@ import { AuthDiagnosticsReadScope, USAGE_CONTRACT_VERSION, type EnvironmentId, + type ScopedThreadRef, type UsageBucket, type UsageSummary, type UsageProviderKind, type UsageSummaryInput, } from "@t3tools/contracts"; -import { refreshUsage, usageSummaryInput } from "@t3tools/client-runtime/state/usage"; +import { parseThreadKey, threadKey } from "@t3tools/client-runtime/state/entities"; +import { + makeTurnEndRefresh, + refreshUsage, + threadUsageSummaryInput, + usageSummaryInput, +} from "@t3tools/client-runtime/state/usage"; import { resolveUsageAccess } from "@t3tools/client-runtime/state/usage-access"; +import * as DateTime from "effect/DateTime"; import { AsyncResult, Atom } from "effect/reactivity"; import { msg } from "gt-react"; import { useState } from "react"; @@ -18,7 +26,8 @@ import { mergeUsage, type EnvironmentUsage, type MergedUsage } from "@t3tools/sh import { appAtomRegistry } from "../rpc/atomRegistry"; import { environmentPresentations } from "./presentation"; import { serverEnvironment } from "./server"; -import { environmentSession, readEnvironmentScope } from "./session"; +import { environmentSession, readEnvironmentScope, sessionHasScope } from "./session"; +import { environmentThreadDetails } from "./threads"; import { m } from "../i18n/messages"; export interface EnvironmentUsageStatus { @@ -247,3 +256,64 @@ export function useUsage( refresh, }; } + +const threadUsageAtom = Atom.family((key: string) => { + const ref = parseThreadKey(key); + const { environmentId, threadId } = ref; + const shouldRefresh = makeTurnEndRefresh(); + // Stays on screen while a changed query, such as a new day's, is still being read. + let answered: MergedUsage | null = null; + return Atom.make((get): MergedUsage | null => { + if ( + !sessionHasScope( + get(environmentSession.sessionStateAtom(environmentId)), + AuthDiagnosticsReadScope, + ) + ) { + return null; + } + const projection = get(environmentThreadDetails.quietThreadAtom(ref))?.projection; + const input = + projection === undefined + ? null + : threadUsageSummaryInput({ + threadId, + createdAt: projection.thread.createdAt, + capabilities: get(environmentPresentations.presentationAtom(environmentId)) + ?.serverConfig?.environment.capabilities, + }); + if (projection === undefined || input === null) return null; + const query = serverEnvironment.usageSummary({ + environmentId, + input: usageSummaryInput(input, get(serverEnvironment.usageModelFilterAtom(environmentId))), + }); + const result = get(query); + const lastTurnEndMs = projection.providerTurns.reduce( + (latest, turn) => + turn.completedAt === null + ? latest + : Math.max(latest, DateTime.toEpochMillis(turn.completedAt)), + 0, + ); + if (shouldRefresh(lastTurnEndMs, result.waiting)) get.refresh(query); + // A failed read has no figure; the one before it would pass for current. + if (AsyncResult.isFailure(result) && !AsyncResult.isInterrupted(result)) { + answered = null; + return null; + } + const summary = AsyncResult.getOrElse(result, () => null); + if (summary === null) return answered; + const merged = mergeUsage([{ environmentId, label: "", summary }], USAGE_CONTRACT_VERSION); + // A summary from another contract version merges to nothing, which is not a zero cost. + answered = merged.contractMismatches.length === 0 ? merged : null; + return answered; + }).pipe(Atom.withLabel(`web-usage:thread:${key}`)); +}); + +/** + * One thread's usage so far, priced like the Usage tab. Null until the server answers, and on a + * server that cannot narrow usage to a thread. + */ +export function useThreadUsage(ref: ScopedThreadRef): MergedUsage | null { + return useAtomValue(threadUsageAtom(threadKey(ref))); +} diff --git a/docs/user/usage.md b/docs/user/usage.md index 5160becc2156..36409796f07a 100644 --- a/docs/user/usage.md +++ b/docs/user/usage.md @@ -157,6 +157,15 @@ row to see when it resets and your pace. Turn on **Context window** to also show context use against the model's limit. A tick on the bar marks where the provider compacts automatically, when it reports that point. +Turn on **Cost** to show what the thread's model usage so far would cost at API prices. It is the +same estimate as the Usage tab, narrowed to the thread, so on a subscription it is not a charge. +Hover the row to see the cost by token type. It covers the thread's own conversations, including +subagents the provider runs inside them, but not other threads it started. The row appears once +the provider's usage for the thread is recorded, and updates when a turn ends. If the provider had +to start a fresh conversation in the thread, such as after a failed resume, usage before it is left +out. The row stays hidden when the saved usage does not name the thread's conversation, as with Pi, when your +session cannot read usage, and on a server too old to narrow usage to a thread. + OpenCode Go reports its session, weekly, and monthly allowance when OpenCode runs locally in the environment. T3 cannot report limits for external OpenCode servers because their credentials belong to the remote server. Limits need an OpenCode Go API key, or an OpenCode Console connected diff --git a/packages/client-runtime/src/state/usage.test.ts b/packages/client-runtime/src/state/usage.test.ts index 380e2de731b5..a3f6c2ff9920 100644 --- a/packages/client-runtime/src/state/usage.test.ts +++ b/packages/client-runtime/src/state/usage.test.ts @@ -1,17 +1,24 @@ import { EnvironmentId, + ThreadId, UsageDay, USAGE_CONTRACT_VERSION, type UsageModelFilter, type UsageSummary, type UsageSummaryInput, } from "@t3tools/contracts"; +import * as DateTime from "effect/DateTime"; import * as Effect from "effect/Effect"; import { AsyncResult, Atom, AtomRegistry } from "effect/reactivity"; import { afterEach, describe, expect, it, vi } from "vite-plus/test"; import type { EnvironmentPresentation } from "../connection/presentation.ts"; -import { refreshUsage, refreshUsageLimits } from "./usage.ts"; +import { + makeTurnEndRefresh, + refreshUsage, + refreshUsageLimits, + threadUsageSummaryInput, +} from "./usage.ts"; const input = { sinceDay: UsageDay.make("2026-09-05"), @@ -205,3 +212,56 @@ describe("limits refresh cooldown", () => { expect(newCheck).not.toHaveBeenCalled(); }); }); + +describe("threadUsageSummaryInput", () => { + const threadId = ThreadId.make("thread-1"); + const capabilities = { usageThreadFilter: true }; + const createdAt = DateTime.makeUnsafe("2026-09-03T08:15:00.000Z"); + const now = DateTime.makeUnsafe("2026-09-05T12:00:00Z"); + + it("asks for the thread's usage from the day it was created", () => { + expect(threadUsageSummaryInput({ threadId, createdAt, capabilities, now })).toMatchObject({ + sinceDay: "2026-09-03", + untilDay: "2026-09-05", + threadId, + }); + }); + + it("asks nothing of a server that would answer with every thread's usage", () => { + expect(threadUsageSummaryInput({ threadId, createdAt, capabilities: {}, now })).toBeNull(); + expect( + threadUsageSummaryInput({ threadId, createdAt, capabilities: undefined, now }), + ).toBeNull(); + }); + + it("keeps the creation day in the window when the viewer's clock trails the server's", () => { + expect( + threadUsageSummaryInput({ + threadId, + createdAt, + capabilities, + now: DateTime.makeUnsafe("2026-09-01T12:00:00Z"), + }), + ).toMatchObject({ sinceDay: "2026-09-03", untilDay: "2026-09-03" }); + }); +}); + +describe("makeTurnEndRefresh", () => { + it("reads again when a turn ends after the answer in hand", () => { + const shouldRefresh = makeTurnEndRefresh(); + expect(shouldRefresh(100, true)).toBe(false); + expect(shouldRefresh(100, false)).toBe(false); + expect(shouldRefresh(200, false)).toBe(true); + expect(shouldRefresh(200, true)).toBe(false); + expect(shouldRefresh(200, false)).toBe(false); + }); + + it("follows a read that a turn end overtook with one more", () => { + const shouldRefresh = makeTurnEndRefresh(); + expect(shouldRefresh(100, true)).toBe(false); + // The turn ends while the first read is still scanning. + expect(shouldRefresh(200, true)).toBe(false); + expect(shouldRefresh(200, false)).toBe(true); + expect(shouldRefresh(200, false)).toBe(false); + }); +}); diff --git a/packages/client-runtime/src/state/usage.ts b/packages/client-runtime/src/state/usage.ts index f302ffbdfd85..30ce002c8730 100644 --- a/packages/client-runtime/src/state/usage.ts +++ b/packages/client-runtime/src/state/usage.ts @@ -1,4 +1,12 @@ -import type { EnvironmentId, UsageModelFilter, UsageSummaryInput } from "@t3tools/contracts"; +import type { + EnvironmentId, + ExecutionEnvironmentCapabilities, + ThreadId, + UsageModelFilter, + UsageSummaryInput, +} from "@t3tools/contracts"; +import { makeWindow } from "@t3tools/shared/usageFormat"; +import * as DateTime from "effect/DateTime"; import * as Schema from "effect/Schema"; import type { AtomRegistry } from "effect/reactivity"; @@ -17,6 +25,50 @@ export function usageSummaryInput( return modelFilter === undefined ? input : { ...input, modelFilter }; } +/** + * The usage query for one thread, from the day it was created. Null for a server that would + * ignore `threadId` and answer with every thread's usage. + */ +export function threadUsageSummaryInput(input: { + readonly threadId: ThreadId; + readonly createdAt: DateTime.Utc; + readonly capabilities: Pick | undefined; + readonly now?: DateTime.Utc; +}): UsageSummaryInput | null { + if (input.capabilities?.usageThreadFilter !== true) return null; + const { untilDay: today, timeZone } = makeWindow( + 1, + input.now === undefined ? undefined : DateTime.toDateUtc(input.now), + ); + const createdDay = makeWindow(1, DateTime.toDateUtc(input.createdAt)).untilDay; + return { + sinceDay: createdDay, + // A viewer whose clock trails the server's can see a thread created "tomorrow". + untilDay: createdDay < today ? today : createdDay, + timeZone, + threadId: input.threadId, + }; +} + +/** + * Decides when to read a thread's usage again. A turn's last records reach the transcript as it + * ends, and a read in flight at that moment may have scanned before them, so that read's answer + * is followed by another. + */ +export function makeTurnEndRefresh(): (lastTurnEndMs: number, waiting: boolean) => boolean { + // The latest turn end that the read in flight, or the answer in hand, began after. + let readAfter: number | null = null; + return (lastTurnEndMs, waiting) => { + if (readAfter === null) { + readAfter = lastTurnEndMs; + return false; + } + if (waiting || readAfter === lastTurnEndMs) return false; + readAfter = lastTurnEndMs; + return true; + }; +} + const limitsRefreshAfter = new Map(); const limitsRefreshes = new Map>(); diff --git a/packages/contracts/src/environment.ts b/packages/contracts/src/environment.ts index b1b014cbc131..e49ff9d782a4 100644 --- a/packages/contracts/src/environment.ts +++ b/packages/contracts/src/environment.ts @@ -132,6 +132,8 @@ export const ExecutionEnvironmentCapabilities = Schema.Struct({ modelCatalogRefresh: Schema.optionalKey(Schema.Boolean), usagePriceOverrides: Schema.optionalKey(Schema.Boolean), usageModelAliases: Schema.optionalKey(Schema.Boolean), + /** Usage summaries honor `threadId`; older servers ignore it and return every thread's usage. */ + usageThreadFilter: Schema.optionalKey(Schema.Boolean), threadPinning: Schema.optionalKey(Schema.Boolean), threadPinReorder: Schema.optionalKey(Schema.Boolean), threadActiveReorder: Schema.optionalKey(Schema.Boolean), diff --git a/packages/contracts/src/usage.ts b/packages/contracts/src/usage.ts index 427d6c93af5c..b0486850d881 100644 --- a/packages/contracts/src/usage.ts +++ b/packages/contracts/src/usage.ts @@ -1,6 +1,11 @@ import * as Schema from "effect/Schema"; -import { ForwardCompatibleArray, NonNegativeInt, TrimmedNonEmptyString } from "./baseSchemas.ts"; +import { + ForwardCompatibleArray, + NonNegativeInt, + ThreadId, + TrimmedNonEmptyString, +} from "./baseSchemas.ts"; /** Bump for incompatible UsageSummary changes; skipped unknown array variants need no bump. */ export const USAGE_CONTRACT_VERSION = 6 as const; @@ -145,6 +150,8 @@ export const UsageSummaryInput = Schema.Struct({ /** Accept an immediate answer from the last scan; its sources are marked refreshing. */ provisional: Schema.optional(Schema.Boolean), modelFilter: Schema.optional(UsageModelFilter), + /** Narrows usage to one thread: its provider sessions, from when it was created. */ + threadId: Schema.optional(ThreadId), }); export type UsageSummaryInput = typeof UsageSummaryInput.Type;