diff --git a/docs/changelog/2026-09-26-gravacoes-sem-teto-4gb.md b/docs/changelog/2026-09-26-gravacoes-sem-teto-4gb.md
new file mode 100644
index 00000000..5adf8481
--- /dev/null
+++ b/docs/changelog/2026-09-26-gravacoes-sem-teto-4gb.md
@@ -0,0 +1 @@
+**Gravações enviadas ao acervo sem o teto de 4 GB (2026-09-26):** o diálogo "Enviar gravação" recusava qualquer arquivo acima de 4 GiB (`RECORDING_MAX_BYTES`) — limite que plenárias longas encostam. O teto saiu do cliente, da rota e do stream: o corpo continua indo para o temp com backpressure, agora sem teto de produto (disco e multipart do S3 são o limite). Acima de 2 GiB a mídia deixa de passar pelo `fs.readFile` do Payload: o `recordingMedia` nasce de um placeholder de 0 byte e a mesma key é sobrescrita por upload multipart direto no S3 — padrão do C225-large, extraído para o dono `utilities/privateMedia/privateMediaUpload.ts` e agora compartilhado com o espelho de falas web (que mantém o próprio teto de 5 GiB via `maxBytes` do chamador; gravações passam `Infinity`). O nome original do arquivo sobrevive ao placeholder e o objeto é removido se a transação falhar, sem row fantasma. Sem migration e sem schema; unit (novo helper + limites do uploader) e int de gravações/falas web verdes. Operação: o `/tmp` do container precisa de headroom para o tamanho da gravação (upload + download do job de transcrição).
diff --git a/src/app/(campaign)/campanha/(app)/comunicacao/acervo/gravacoes/enviar/route.ts b/src/app/(campaign)/campanha/(app)/comunicacao/acervo/gravacoes/enviar/route.ts
index 4eb882f0..b9b7cd26 100644
--- a/src/app/(campaign)/campanha/(app)/comunicacao/acervo/gravacoes/enviar/route.ts
+++ b/src/app/(campaign)/campanha/(app)/comunicacao/acervo/gravacoes/enviar/route.ts
@@ -7,7 +7,6 @@ import {
   RECORDING_FILE_TYPE_MESSAGE,
   RECORDING_TITLE_REQUIRED_MESSAGE,
   recordingFileTypeAllowed,
-  recordingUploadTooLargeMessage,
   toRecordingViewModel,
 } from '@/lib/recording'
 import {
@@ -49,7 +48,6 @@ const SAFE_MESSAGES = [
   RECORDING_TITLE_REQUIRED_MESSAGE,
   RECORDING_FILE_NAME_MESSAGE,
   RECORDING_FILE_TYPE_MESSAGE,
-  recordingUploadTooLargeMessage,
   RECORDING_BODY_MISSING_MESSAGE,
 ]
 
@@ -78,13 +76,6 @@ export const POST = async (request: Request): Promise<NextResponse<RecordingUplo
     return errorResponse(RECORDING_FILE_TYPE_MESSAGE, 400)
   }
 
-  const contentLengthHeader = request.headers.get('content-length')
-  const parsedLength = contentLengthHeader === null ? null : Number(contentLengthHeader)
-  const contentLength =
-    parsedLength !== null && Number.isFinite(parsedLength) && parsedLength >= 0
-      ? parsedLength
-      : null
-
   try {
     const payload = await getPayload({ config })
     const { id } = await receiveRecordingUpload({
@@ -92,7 +83,6 @@ export const POST = async (request: Request): Promise<NextResponse<RecordingUplo
       actor: user,
       metadata: parsed.data,
       body: request.body,
-      contentLength,
     })
 
     const recording = await payload.findByID({
diff --git a/src/components/campaign/recording/RecordingUploadDialog.tsx b/src/components/campaign/recording/RecordingUploadDialog.tsx
index 7998d810..4febb7eb 100644
--- a/src/components/campaign/recording/RecordingUploadDialog.tsx
+++ b/src/components/campaign/recording/RecordingUploadDialog.tsx
@@ -24,12 +24,9 @@ import {
   formatRecordingFileSize,
   RECORDING_FILE_REQUIRED_MESSAGE,
   RECORDING_FILE_TYPE_MESSAGE,
-  RECORDING_MAX_BYTES,
-  RECORDING_MAX_SIZE_LABEL,
   RECORDING_TITLE_MAX_LENGTH,
   RECORDING_TITLE_REQUIRED_MESSAGE,
   recordingFileTypeAllowed,
-  recordingTooLargeMessage,
 } from '@/lib/recording'
 import { cn } from '@/lib/utils'
 
@@ -109,10 +106,6 @@ export const RecordingUploadDialog = ({
     setFile(next)
     if (!recordingFileTypeAllowed(next.type)) {
       setFileError(RECORDING_FILE_TYPE_MESSAGE)
-      return
-    }
-    if (next.size > RECORDING_MAX_BYTES) {
-      setFileError(recordingTooLargeMessage(next.size))
     }
   }
 
@@ -285,7 +278,7 @@ export const RecordingUploadDialog = ({
                     <UploadIcon className="size-6 text-muted-foreground" aria-hidden="true" />
                     <span className="mt-2 text-sm font-medium">Selecionar arquivo local</span>
                     <span className="mt-1 text-xs text-muted-foreground">
-                      Vídeos até {RECORDING_MAX_SIZE_LABEL} · MP4, MOV, MKV e formatos compatíveis
+                      MP4, MOV, MKV e formatos compatíveis
                     </span>
                   </>
                 )}
@@ -297,7 +290,7 @@ export const RecordingUploadDialog = ({
                     <AlertDescription className="text-xs">{fileError}</AlertDescription>
                   </Alert>
                   <p className="mt-2 text-xs text-muted-foreground">
-                    Vídeos até {RECORDING_MAX_SIZE_LABEL} · MP4, MOV, MKV e formatos compatíveis
+                    MP4, MOV, MKV e formatos compatíveis
                   </p>
                 </>
               ) : null}
diff --git a/src/lib/recording.ts b/src/lib/recording.ts
index eed402a4..bed63a57 100644
--- a/src/lib/recording.ts
+++ b/src/lib/recording.ts
@@ -1,7 +1,7 @@
 /**
- * C199 — pure rules of an uploaded recording: the status/step vocabulary, the
- * upload limits and the honest failure copy. No I/O and no `server-only`: the
- * client dialog, the JSON routes, the job and the unit tests share this module.
+ * C199 — pure rules of an uploaded recording: the status/step vocabulary and
+ * the honest failure copy. No I/O and no `server-only`: the client dialog, the
+ * JSON routes, the job and the unit tests share this module.
  */
 import {
   CAMPAIGN_COMMUNICATION_ACERVO_GRAVACOES,
@@ -35,9 +35,7 @@ export const RECORDING_SPEAKER_LABEL_MAX_LENGTH = 120
 export const RECORDING_SPEAKER_LABEL_REQUIRED_MESSAGE = 'Informe o nome do falante.'
 export const RECORDING_SPEAKER_LABEL_LONG_MESSAGE = `O nome do falante deve ter até ${RECORDING_SPEAKER_LABEL_MAX_LENGTH} caracteres.`
 
-/** Upload ceiling: hours of plenary are the real case, 4 GiB is the guard. */
-export const RECORDING_MAX_BYTES = 4 * 1024 * 1024 * 1024
-export const RECORDING_MAX_SIZE_LABEL = '4 GB'
+/** Accepted upload type: any video container (`.mkv` often arrives with no type). */
 const RECORDING_ACCEPTED_MIME_PREFIX = 'video/'
 
 export const RECORDING_STEPS = ['extracting', 'transcribing', 'saving'] as const
@@ -147,11 +145,6 @@ export const formatRecordingFileSize = (bytes: number): string => {
   return `${Math.round(bytes)} B`
 }
 
-export const recordingTooLargeMessage = (bytes: number): string =>
-  `Este arquivo tem ${formatRecordingFileSize(bytes)}. Escolha um vídeo de até ${RECORDING_MAX_SIZE_LABEL}.`
-
-export const recordingUploadTooLargeMessage = `O arquivo excede o limite de ${RECORDING_MAX_SIZE_LABEL}.`
-
 export type RecordingViewModel = {
   id: number
   title: string
diff --git a/src/utilities/media/largeS3Upload.ts b/src/utilities/media/largeS3Upload.ts
index 266609c1..a185d1ce 100644
--- a/src/utilities/media/largeS3Upload.ts
+++ b/src/utilities/media/largeS3Upload.ts
@@ -17,7 +17,10 @@ const MIME_BY_EXTENSION: Record<string, string> = {
   '.aac': 'audio/aac',
   '.flac': 'audio/flac',
   '.m4a': 'audio/mp4',
+  '.m4v': 'video/x-m4v',
   '.mka': 'audio/x-matroska',
+  '.mkv': 'video/x-matroska',
+  '.mov': 'video/quicktime',
   '.mp3': 'audio/mpeg',
   '.mp4': 'video/mp4',
   '.ogg': 'audio/ogg',
@@ -116,18 +119,21 @@ export const uploadLargeMedia = async ({
   inputPath,
   filename: providedFilename,
   env = process.env,
+  maxBytes = MAX_LARGE_MEDIA_BYTES,
   upload = defaultUpload,
 }: {
   inputPath: string
   filename?: string
   env?: Record<string, string | undefined>
+  /** Caller ceiling; `Infinity` drops it (the S3 multipart limit is the bound). */
+  maxBytes?: number
   upload?: (args: S3UploadArgs) => Promise<void>
 }): Promise<LargeMediaUpload> => {
   const storage = requireStorage(env)
   const { size } = await stat(inputPath)
-  if (size > MAX_LARGE_MEDIA_BYTES) {
+  if (size > maxBytes) {
     throw new Error(
-      `Mídia de ${(size / 1024 / 1024 / 1024).toFixed(2)} GiB excede o limite de 5 GiB.`,
+      `Mídia de ${(size / 1024 / 1024 / 1024).toFixed(2)} GiB excede o limite de ${(maxBytes / 1024 / 1024 / 1024).toFixed(0)} GiB.`,
     )
   }
 
diff --git a/src/utilities/privateMedia/privateMediaUpload.ts b/src/utilities/privateMedia/privateMediaUpload.ts
new file mode 100644
index 00000000..6aa527d0
--- /dev/null
+++ b/src/utilities/privateMedia/privateMediaUpload.ts
@@ -0,0 +1,95 @@
+import 'server-only'
+
+import { mkdir, stat, writeFile } from 'node:fs/promises'
+import { dirname, extname, join } from 'node:path'
+
+import {
+  deleteLargeMedia,
+  isLargeMedia,
+  LARGE_MEDIA_THRESHOLD_BYTES,
+  uploadLargeMedia,
+} from '@/utilities/media/largeS3Upload'
+
+/**
+ * C225-large/C199 — creates one private upload row from a local file without
+ * ever letting Payload's local API `fs.readFile` an arbitrarily large artifact:
+ * above the buffer boundary the row is created through an empty placeholder and
+ * the same object key is overwritten by a direct multipart stream, then the
+ * stored `filesize`/MIME metadata is updated. Below the boundary the Payload
+ * path is used as-is. The caller's `create`/`update` closures own collection,
+ * access and transaction details.
+ *
+ * `cleanup` removes the multipart object; call it when the surrounding
+ * transaction fails, since the row is rolled back with it.
+ */
+
+export type PrivateMediaFromFile = {
+  id: number
+  cleanup: () => Promise<void>
+}
+
+type PrivateMediaRow = { id: number; filename?: string | null }
+
+export type CreatePrivateMediaFromFileArgs = {
+  inputPath: string
+  /**
+   * Placeholder file the large path creates the row from; its basename is the
+   * stored name Payload derives. Override it to keep the original file name.
+   */
+  placeholderPath?: string
+  create: (filePath: string) => Promise<PrivateMediaRow>
+  update: (id: number, data: { filesize: number; mimeType: string }) => Promise<unknown>
+  thresholdBytes?: number
+  /** Multipart ceiling; omit to keep the uploader default. */
+  maxBytes?: number
+  uploadLarge?: typeof uploadLargeMedia
+  removeLarge?: typeof deleteLargeMedia
+}
+
+const noopCleanup = async (): Promise<void> => {}
+
+export const createPrivateMediaFromFile = async ({
+  inputPath,
+  placeholderPath,
+  create,
+  update,
+  thresholdBytes = LARGE_MEDIA_THRESHOLD_BYTES,
+  maxBytes,
+  uploadLarge = uploadLargeMedia,
+  removeLarge = deleteLargeMedia,
+}: CreatePrivateMediaFromFileArgs): Promise<PrivateMediaFromFile> => {
+  const { size } = await stat(inputPath)
+  if (!isLargeMedia(size, thresholdBytes)) {
+    const media = await create(inputPath)
+    return { id: media.id, cleanup: noopCleanup }
+  }
+
+  const placeholder =
+    placeholderPath ?? join(dirname(inputPath), `payload-placeholder${extname(inputPath)}`)
+  await mkdir(dirname(placeholder), { recursive: true })
+  await writeFile(placeholder, Buffer.alloc(0))
+
+  let uploadedFilename: string | null = null
+  try {
+    const media = await create(placeholder)
+    uploadedFilename = media.filename ?? null
+    if (!uploadedFilename) throw new Error('A mídia grande não recebeu um filename no S3.')
+    const uploaded = await uploadLarge({
+      inputPath,
+      filename: uploadedFilename,
+      ...(maxBytes === undefined ? {} : { maxBytes }),
+    })
+    await update(media.id, { filesize: uploaded.filesize, mimeType: uploaded.mimeType })
+
+    const filename = uploadedFilename
+    return {
+      id: media.id,
+      cleanup: async () => {
+        await removeLarge({ filename }).catch(() => undefined)
+      },
+    }
+  } catch (error) {
+    if (uploadedFilename) await removeLarge({ filename: uploadedFilename }).catch(() => undefined)
+    throw error
+  }
+}
diff --git a/src/utilities/recordings/recordingUpload.ts b/src/utilities/recordings/recordingUpload.ts
index 3f0a9eba..7fae3893 100644
--- a/src/utilities/recordings/recordingUpload.ts
+++ b/src/utilities/recordings/recordingUpload.ts
@@ -4,65 +4,45 @@ import { createWriteStream } from 'node:fs'
 import { mkdtemp, rm } from 'node:fs/promises'
 import { tmpdir } from 'node:os'
 import { join } from 'node:path'
-import { Readable, Transform } from 'node:stream'
+import { Readable } from 'node:stream'
 import { pipeline } from 'node:stream/promises'
 import type { Payload } from 'payload'
 
-import {
-  RECORDING_MAX_BYTES,
-  RECORDING_MEDIA_SLUG,
-  recordingUploadTooLargeMessage,
-  sanitizeRecordingFilename,
-} from '@/lib/recording'
+import { RECORDING_MEDIA_SLUG, sanitizeRecordingFilename } from '@/lib/recording'
 import type { RecordingUploadMetadata } from '@/lib/schemas/recording'
 import type { CampaignUser } from '@/payload-types'
+import { deleteLargeMedia, uploadLargeMedia } from '@/utilities/media/largeS3Upload'
 import { withPayloadTransaction } from '@/utilities/payloadTransaction'
+import { createPrivateMediaFromFile } from '@/utilities/privateMedia/privateMediaUpload'
 import { startRecordingJobInBackground } from '@/utilities/recordings/recordingScheduler'
 
 /**
  * C199 — the upload half of the recordings flow: create the visible row, stream
- * the raw request body to a temp file with the size ceiling enforced, then store
- * the private media and hand the row to the transcription job. The route owns
- * the origin/session/role gate; this module owns the write path.
+ * the raw request body to a temp file, then store the private media (multipart
+ * above the Payload buffer boundary, see `privateMediaUpload`) and hand the row
+ * to the transcription job. The route owns the origin/session/role gate; this
+ * module owns the write path.
  */
 
-/** Shared literal: the route allowlists it and the stream guard throws it. */
+/** Shared literal: the route allowlists it and the missing body throws it. */
 export const RECORDING_BODY_MISSING_MESSAGE = 'A requisição não trouxe o arquivo da gravação.'
 
-const toUploadTooLarge = (): Error => new Error(recordingUploadTooLargeMessage)
-
 /**
- * Streams the request body to `destination`, enforcing the ceiling even when
- * the request arrives chunked (a missing/lying `Content-Length` cannot fill the
- * disk). Returns the received byte count.
+ * Streams the request body to `destination` with backpressure. There is no
+ * product ceiling (the upload accepts hours of plenary): the container disk and
+ * the S3 multipart limit are the bounds.
  */
 const streamBodyToFile = async ({
   body,
   destination,
-  maxBytes,
 }: {
   body: ReadableStream<Uint8Array>
   destination: string
-  maxBytes: number
-}): Promise<number> => {
-  let received = 0
-  const guard = new Transform({
-    transform(chunk: Buffer, _encoding, callback) {
-      received += chunk.length
-      if (received > maxBytes) {
-        callback(toUploadTooLarge())
-        return
-      }
-      callback(null, chunk)
-    },
-  })
-
+}): Promise<void> => {
   await pipeline(
     Readable.fromWeb(body as unknown as import('node:stream/web').ReadableStream),
-    guard,
     createWriteStream(destination),
   )
-  return received
 }
 
 export const receiveRecordingUpload = async ({
@@ -70,18 +50,22 @@ export const receiveRecordingUpload = async ({
   actor,
   metadata,
   body,
-  contentLength,
   startJob = startRecordingJobInBackground,
+  largeThresholdBytes,
+  uploadLarge,
+  removeLarge,
 }: {
   payload: Payload
   actor: CampaignUser
   metadata: RecordingUploadMetadata
   body: ReadableStream<Uint8Array> | null
-  contentLength: number | null
   /** Injectable for tests; the route uses the `after()`-based default. */
   startJob?: (recordingId: number) => void
+  /** Test seams; the production defaults live in the private-media owner. */
+  largeThresholdBytes?: number
+  uploadLarge?: typeof uploadLargeMedia
+  removeLarge?: typeof deleteLargeMedia
 }): Promise<{ id: number }> => {
-  if (contentLength !== null && contentLength > RECORDING_MAX_BYTES) throw toUploadTooLarge()
   if (!body) throw new Error(RECORDING_BODY_MISSING_MESSAGE)
 
   const recording = await payload.create({
@@ -97,22 +81,48 @@ export const receiveRecordingUpload = async ({
   })
 
   let tempDir: string | null = null
+  const mediaCleanup: { current?: () => Promise<void> } = {}
   try {
-    tempDir = await mkdtemp(join(tmpdir(), 'recording-upload-'))
+    const uploadDir = await mkdtemp(join(tmpdir(), 'recording-upload-'))
+    tempDir = uploadDir
     const uploadName = sanitizeRecordingFilename(metadata.filename)
-    const tempPath = join(tempDir, uploadName)
-    await streamBodyToFile({ body, destination: tempPath, maxBytes: RECORDING_MAX_BYTES })
+    const tempPath = join(uploadDir, uploadName)
+    await streamBodyToFile({ body, destination: tempPath })
 
     await withPayloadTransaction(payload, async ({ req }) => {
-      const media = await payload.create({
-        collection: RECORDING_MEDIA_SLUG,
-        data: { alt: metadata.title },
-        filePath: tempPath,
-        depth: 0,
-        user: actor,
-        overrideAccess: false,
-        req,
+      const media = await createPrivateMediaFromFile({
+        inputPath: tempPath,
+        // Same basename as the upload: the stored name (and the download) keeps
+        // the original file name even on the multipart path.
+        placeholderPath: join(uploadDir, 'placeholder', uploadName),
+        thresholdBytes: largeThresholdBytes,
+        // No product ceiling for recordings; disk is the bound.
+        maxBytes: Number.POSITIVE_INFINITY,
+        uploadLarge,
+        removeLarge,
+        create: (filePath) =>
+          payload.create({
+            collection: RECORDING_MEDIA_SLUG,
+            data: { alt: metadata.title },
+            filePath,
+            depth: 0,
+            user: actor,
+            overrideAccess: false,
+            req,
+          }),
+        update: (id, data) =>
+          payload.update({
+            collection: RECORDING_MEDIA_SLUG,
+            id,
+            data,
+            depth: 0,
+            user: actor,
+            overrideAccess: false,
+            req,
+          }),
       })
+      mediaCleanup.current = media.cleanup
+
       await payload.update({
         collection: 'recording',
         id: recording.id,
@@ -123,12 +133,16 @@ export const receiveRecordingUpload = async ({
         req,
       })
     })
+    // Committed: the multipart object is the recording's media now.
+    mediaCleanup.current = undefined
 
     startJob(recording.id)
     return { id: recording.id }
   } catch (error) {
     // The upload never completed: the row must not linger as a phantom
-    // "Enviando". The media (when it exists) is unreachable without the row.
+    // "Enviando". A rolled-back transaction leaves the multipart object (when
+    // the large path ran) with no row owning it, so it is removed too.
+    await mediaCleanup.current?.().catch(() => undefined)
     await payload
       .delete({
         collection: 'recording',
diff --git a/src/utilities/speech/webSpeechIngest.ts b/src/utilities/speech/webSpeechIngest.ts
index 9643b24e..9f766983 100644
--- a/src/utilities/speech/webSpeechIngest.ts
+++ b/src/utilities/speech/webSpeechIngest.ts
@@ -1,8 +1,8 @@
 import 'server-only'
 
-import { mkdtemp, readdir, readFile, rm, stat, writeFile } from 'node:fs/promises'
+import { mkdtemp, readdir, readFile, rm } from 'node:fs/promises'
 import { tmpdir } from 'node:os'
-import { dirname, extname, join } from 'node:path'
+import { join } from 'node:path'
 import type { Payload } from 'payload'
 
 import {
@@ -32,10 +32,8 @@ import { downloadUrlToFile } from '@/utilities/media/downloadToFile'
 import { ffmpegBinary, messageOf, runFfmpeg } from '@/utilities/media/ffmpeg'
 import {
   deleteLargeMedia,
-  isLargeMedia,
   LARGE_MEDIA_THRESHOLD_BYTES,
   uploadLargeMedia,
-  type LargeMediaUpload,
 } from '@/utilities/media/largeS3Upload'
 import {
   downloadWithYtDlp,
@@ -48,6 +46,7 @@ import {
   withPayloadTransaction,
   type PayloadTransactionRequest,
 } from '@/utilities/payloadTransaction'
+import { createPrivateMediaFromFile } from '@/utilities/privateMedia/privateMediaUpload'
 import {
   classifySpeech,
   type SpeechClassificationResult,
@@ -293,8 +292,6 @@ const extractTranscript = async ({
   return { segments, durationSeconds, asrSeconds }
 }
 
-const noopCleanup = async (): Promise<void> => {}
-
 type WebSpeechMediaMirrorOptions = {
   thresholdBytes?: number
   uploadLarge?: typeof uploadLargeMedia
@@ -307,60 +304,31 @@ export const createWebSpeechMediaMirror =
     uploadLarge = uploadLargeMedia,
     removeLarge = deleteLargeMedia,
   }: WebSpeechMediaMirrorOptions = {}): WebSpeechMediaMirror =>
-  async ({ payload, req, alt, inputPath }) => {
-    const { size } = await stat(inputPath)
-    if (!isLargeMedia(size, thresholdBytes)) {
-      const media = await payload.create({
-        collection: INTERNET_SPEECH_MEDIA_SLUG,
-        data: { alt },
-        filePath: inputPath,
-        req,
-        // Intentional bypass: the ingestion CLI is a trusted actor with no session.
-        overrideAccess: true,
-      })
-      return { id: media.id, cleanup: noopCleanup }
-    }
-
-    const placeholderPath = join(dirname(inputPath), `payload-placeholder${extname(inputPath)}`)
-    await writeFile(placeholderPath, Buffer.alloc(0))
-    let filename: string | null = null
-
-    try {
-      const media = await payload.create({
-        collection: INTERNET_SPEECH_MEDIA_SLUG,
-        data: { alt },
-        filePath: placeholderPath,
-        req,
-        // Intentional bypass: the placeholder row is owned by the trusted ingestion CLI.
-        overrideAccess: true,
-      })
-      filename = media.filename ?? null
-      if (!filename) throw new Error('A mídia grande não recebeu um filename no S3.')
-      const uploadedFilename = filename
-
-      const uploaded: LargeMediaUpload = await uploadLarge({
-        inputPath,
-        filename: uploadedFilename,
-      })
-      await payload.update({
-        collection: INTERNET_SPEECH_MEDIA_SLUG,
-        id: media.id,
-        data: { filesize: uploaded.filesize, mimeType: uploaded.mimeType },
-        req,
-        // Intentional bypass: the trusted ingestion CLI finalizes its own media row.
-        overrideAccess: true,
-      })
-      return {
-        id: media.id,
-        cleanup: async () => {
-          await removeLarge({ filename: uploadedFilename }).catch(() => undefined)
-        },
-      }
-    } catch (error) {
-      if (filename) await removeLarge({ filename }).catch(() => undefined)
-      throw error
-    }
-  }
+  ({ payload, req, alt, inputPath }) =>
+    createPrivateMediaFromFile({
+      inputPath,
+      thresholdBytes,
+      uploadLarge,
+      removeLarge,
+      create: (filePath) =>
+        payload.create({
+          collection: INTERNET_SPEECH_MEDIA_SLUG,
+          data: { alt },
+          filePath,
+          req,
+          // Intentional bypass: the ingestion CLI is a trusted actor with no session.
+          overrideAccess: true,
+        }),
+      update: (id, data) =>
+        payload.update({
+          collection: INTERNET_SPEECH_MEDIA_SLUG,
+          id,
+          data,
+          req,
+          // Intentional bypass: the trusted ingestion CLI finalizes its own media row.
+          overrideAccess: true,
+        }),
+    })
 
 const defaultMirrorMedia = createWebSpeechMediaMirror()
 
diff --git a/tests/int/recording.int.spec.ts b/tests/int/recording.int.spec.ts
index fdcf722b..fd6d6c7f 100644
--- a/tests/int/recording.int.spec.ts
+++ b/tests/int/recording.int.spec.ts
@@ -33,7 +33,6 @@ import {
   retryRecordingForActor,
 } from '@/app/(campaign)/campanha/actions/recording'
 import { getMunicipalityCatalogEntry } from '@/lib/municipalityCatalog'
-import { RECORDING_MAX_BYTES } from '@/lib/recording'
 import {
   RECORDING_FORBIDDEN_MESSAGE,
   RECORDING_SPEAKER_UNKNOWN_MESSAGE,
@@ -345,7 +344,6 @@ describe('uploaded recordings (C199)', () => {
         recordedAt: undefined,
       },
       body,
-      contentLength: bytes.length,
       startJob: (recordingId) => scheduled.push(recordingId),
     })
     createdRecordingIds.add(id)
@@ -375,31 +373,97 @@ describe('uploaded recordings (C199)', () => {
     expect(Buffer.from(await response.arrayBuffer())).toEqual(bytes)
   })
 
-  it('refuses an upload above the ceiling without leaving a phantom row', async () => {
+  it('accepts a recording above the payload buffer boundary through the multipart seam', async () => {
+    const bytes = Buffer.from('plenaria-de-horas')
     const body = new ReadableStream<Uint8Array>({
       start(controller) {
-        controller.enqueue(Buffer.alloc(8))
+        controller.enqueue(bytes)
         controller.close()
       },
     })
+    const scheduled: number[] = []
+    const uploaded: { filename?: string; filesize: number }[] = []
+    const removed: string[] = []
+
+    const { id } = await receiveRecordingUpload({
+      payload,
+      actor: communicator,
+      metadata: { title: 'Sessão longa', filename: 'sessao longa.MP4', recordedAt: undefined },
+      body,
+      largeThresholdBytes: 0,
+      startJob: (recordingId) => scheduled.push(recordingId),
+      uploadLarge: async ({ filename }) => {
+        uploaded.push({ filename, filesize: bytes.length })
+        return {
+          filename: filename ?? 'sessao_longa.MP4',
+          mimeType: 'video/mp4',
+          filesize: bytes.length,
+          url: 'http://127.0.0.1:3900/teqo-media/sessao_longa.MP4',
+        }
+      },
+      removeLarge: async ({ filename }) => {
+        removed.push(filename)
+      },
+    })
+    createdRecordingIds.add(id)
+
+    const recording = await payload.findByID({
+      collection: 'recording',
+      id,
+      depth: 1,
+      overrideAccess: true,
+    })
+    expect(recording.status).toBe('processing')
+    expect(recording.step).toBe('extracting')
+    expect(scheduled).toEqual([id])
+
+    const media = recording.media
+    expect(media && typeof media === 'object').toBe(true)
+    if (media && typeof media === 'object') {
+      createdMediaIds.add(media.id)
+      // The original file name survives the placeholder (never the placeholder name).
+      expect(media.filename).toContain('sessao_longa')
+      expect(media.filesize).toBe(bytes.length)
+      expect(media.mimeType).toBe('video/mp4')
+      expect(uploaded).toEqual([{ filename: media.filename, filesize: bytes.length }])
+    }
+    expect(removed).toEqual([])
+  })
+
+  it('removes the multipart object and leaves no phantom row when the large upload fails', async () => {
+    const body = new ReadableStream<Uint8Array>({
+      start(controller) {
+        controller.enqueue(Buffer.from('falha-grande'))
+        controller.close()
+      },
+    })
+    const removed: string[] = []
 
     await expect(
       receiveRecordingUpload({
         payload,
         actor: communicator,
-        metadata: { title: 'Grande demais', filename: 'grande.mp4', recordedAt: undefined },
+        metadata: { title: 'Grande que falha', filename: 'falha.mp4', recordedAt: undefined },
         body,
-        contentLength: RECORDING_MAX_BYTES + 1,
+        largeThresholdBytes: 0,
         startJob: () => undefined,
+        uploadLarge: async () => {
+          throw new Error('S3 indisponível.')
+        },
+        removeLarge: async ({ filename }) => {
+          removed.push(filename)
+        },
       }),
-    ).rejects.toThrow('O arquivo excede o limite de 4 GB.')
+    ).rejects.toThrow('S3 indisponível.')
 
     const rows = await payload.find({
       collection: 'recording',
-      where: { title: { equals: 'Grande demais' } },
+      where: { title: { equals: 'Grande que falha' } },
       overrideAccess: true,
     })
     expect(rows.docs).toHaveLength(0)
+    expect(removed).toHaveLength(1)
+    expect(removed[0]).toContain('falha')
   })
 
   it('preserves a long recording transcript and its search text', async () => {
diff --git a/tests/unit/largeS3Upload.unit.spec.ts b/tests/unit/largeS3Upload.unit.spec.ts
index 9c6c14d3..1e0c184e 100644
--- a/tests/unit/largeS3Upload.unit.spec.ts
+++ b/tests/unit/largeS3Upload.unit.spec.ts
@@ -40,9 +40,36 @@ describe('large S3 media upload', () => {
 
   it('maps the media extension to a safe content type', () => {
     expect(mimeTypeForPath('/tmp/source.mp4')).toBe('video/mp4')
+    expect(mimeTypeForPath('/tmp/source.mov')).toBe('video/quicktime')
+    expect(mimeTypeForPath('/tmp/source.mkv')).toBe('video/x-matroska')
+    expect(mimeTypeForPath('/tmp/source.m4v')).toBe('video/x-m4v')
     expect(mimeTypeForPath('/tmp/source.unknown')).toBe('application/octet-stream')
   })
 
+  it('honors the default ceiling and an explicit caller override', async () => {
+    await withTempFile(async (path) => {
+      await expect(
+        uploadLargeMedia({
+          inputPath: path,
+          env: s3Env,
+          maxBytes: 8,
+          upload: async () => undefined,
+        }),
+      ).rejects.toThrow('excede o limite')
+
+      const keys: string[] = []
+      await uploadLargeMedia({
+        inputPath: path,
+        env: s3Env,
+        maxBytes: Number.POSITIVE_INFINITY,
+        upload: async ({ key }) => {
+          keys.push(key)
+        },
+      })
+      expect(keys).toHaveLength(1)
+    })
+  })
+
   it('streams a large file to S3 and returns row metadata', async () => {
     await withTempFile(async (path) => {
       const calls: { key: string; contentType: string; filesize: number }[] = []
diff --git a/tests/unit/privateMediaUpload.unit.spec.ts b/tests/unit/privateMediaUpload.unit.spec.ts
new file mode 100644
index 00000000..07c07af0
--- /dev/null
+++ b/tests/unit/privateMediaUpload.unit.spec.ts
@@ -0,0 +1,140 @@
+// @vitest-environment node
+
+import { mkdtemp, readFile, rm, stat, writeFile } from 'node:fs/promises'
+import { tmpdir } from 'node:os'
+import { join } from 'node:path'
+
+import { describe, expect, it } from 'vitest'
+
+import { createPrivateMediaFromFile } from '@/utilities/privateMedia/privateMediaUpload'
+
+const withTempFile = async (
+  bytes: Buffer,
+  run: (path: string, dir: string) => Promise<void>,
+): Promise<void> => {
+  const dir = await mkdtemp(join(tmpdir(), 'private-media-upload-test-'))
+  try {
+    const path = join(dir, 'source-video.mp4')
+    await writeFile(path, bytes)
+    await run(path, dir)
+  } finally {
+    await rm(dir, { recursive: true, force: true })
+  }
+}
+
+describe('private media from a local file', () => {
+  it('uses the Payload path below the buffer boundary', async () => {
+    await withTempFile(Buffer.alloc(16), async (path) => {
+      const created: string[] = []
+      let uploads = 0
+      const result = await createPrivateMediaFromFile({
+        inputPath: path,
+        thresholdBytes: 1024,
+        create: async (filePath) => {
+          created.push(filePath)
+          return { id: 7, filename: 'source-video.mp4' }
+        },
+        update: async () => undefined,
+        uploadLarge: async () => {
+          uploads += 1
+          throw new Error('não deveria subir em multipart')
+        },
+      })
+
+      expect(result.id).toBe(7)
+      expect(created).toEqual([path])
+      expect(uploads).toBe(0)
+      await result.cleanup()
+    })
+  })
+
+  it('creates the row from a placeholder and overwrites it through the multipart seam', async () => {
+    await withTempFile(Buffer.alloc(16), async (path, dir) => {
+      const created: string[] = []
+      const updated: { id: number; data: { filesize: number; mimeType: string } }[] = []
+      const uploaded: { inputPath: string; filename?: string; maxBytes?: number }[] = []
+      const removed: string[] = []
+
+      const result = await createPrivateMediaFromFile({
+        inputPath: path,
+        placeholderPath: join(dir, 'placeholder', 'meu-video.mp4'),
+        thresholdBytes: 0,
+        maxBytes: Number.POSITIVE_INFINITY,
+        create: async (filePath) => {
+          created.push(filePath)
+          return { id: 42, filename: 'meu-video.mp4' }
+        },
+        update: async (id, data) => {
+          updated.push({ id, data })
+          return { id }
+        },
+        uploadLarge: async (args) => {
+          uploaded.push(args)
+          return {
+            filename: args.filename ?? 'meu-video.mp4',
+            mimeType: 'video/mp4',
+            filesize: 16,
+            url: 'http://127.0.0.1:3900/teqo-media/meu-video.mp4',
+          }
+        },
+        removeLarge: async ({ filename }) => {
+          removed.push(filename)
+        },
+      })
+
+      expect(result.id).toBe(42)
+      expect(created[0]).toContain(join('placeholder', 'meu-video.mp4'))
+      expect((await stat(created[0])).size).toBe(0)
+      expect(await readFile(created[0])).toEqual(Buffer.alloc(0))
+      expect(uploaded).toEqual([
+        { inputPath: path, filename: 'meu-video.mp4', maxBytes: Number.POSITIVE_INFINITY },
+      ])
+      expect(updated).toEqual([{ id: 42, data: { filesize: 16, mimeType: 'video/mp4' } }])
+      expect(removed).toEqual([])
+
+      await result.cleanup()
+      expect(removed).toEqual(['meu-video.mp4'])
+    })
+  })
+
+  it('removes the object when the multipart upload fails', async () => {
+    await withTempFile(Buffer.alloc(16), async (path) => {
+      const removed: string[] = []
+      await expect(
+        createPrivateMediaFromFile({
+          inputPath: path,
+          thresholdBytes: 0,
+          create: async () => ({ id: 9, filename: 'falha.mp4' }),
+          update: async () => undefined,
+          uploadLarge: async () => {
+            throw new Error('S3 indisponível.')
+          },
+          removeLarge: async ({ filename }) => {
+            removed.push(filename)
+          },
+        }),
+      ).rejects.toThrow('S3 indisponível.')
+      expect(removed).toEqual(['falha.mp4'])
+    })
+  })
+
+  it('fails closed when the placeholder row has no filename', async () => {
+    await withTempFile(Buffer.alloc(16), async (path) => {
+      let uploads = 0
+      await expect(
+        createPrivateMediaFromFile({
+          inputPath: path,
+          thresholdBytes: 0,
+          create: async () => ({ id: 9, filename: null }),
+          update: async () => undefined,
+          uploadLarge: async () => {
+            uploads += 1
+            throw new Error('não deveria subir')
+          },
+          removeLarge: async () => undefined,
+        }),
+      ).rejects.toThrow('filename')
+      expect(uploads).toBe(0)
+    })
+  })
+})
diff --git a/tests/unit/recording.unit.spec.ts b/tests/unit/recording.unit.spec.ts
index ad1085a2..2bd5de98 100644
--- a/tests/unit/recording.unit.spec.ts
+++ b/tests/unit/recording.unit.spec.ts
@@ -8,15 +8,11 @@ import {
   formatRecordingFileSize,
   isRecordingStatus,
   RECORDING_FAILURE_INTERRUPTED,
-  RECORDING_MAX_BYTES,
-  RECORDING_MAX_SIZE_LABEL,
   RECORDING_STATUSES,
   RECORDING_STEPS,
   recordingFailureMessage,
   recordingFileTypeAllowed,
   recordingStatusLabels,
-  recordingTooLargeMessage,
-  recordingUploadTooLargeMessage,
   toRecordingViewModel,
 } from '@/lib/recording'
 import {
@@ -78,13 +74,9 @@ describe('recording vocabulary (C199)', () => {
     expect(recordingFileTypeAllowed('image/png')).toBe(false)
   })
 
-  it('formats sizes in pt-BR and names the limit', () => {
-    expect(formatRecordingFileSize(4.6 * 1024 ** 3)).toContain('GB')
-    expect(formatRecordingFileSize(812 * 1024 ** 2)).toBe('812 MB')
-    expect(recordingTooLargeMessage(4.6 * 1024 ** 3)).toContain('4,6 GB')
-    expect(recordingTooLargeMessage(4.6 * 1024 ** 3)).toContain(RECORDING_MAX_SIZE_LABEL)
-    expect(recordingUploadTooLargeMessage).toContain(RECORDING_MAX_SIZE_LABEL)
-    expect(RECORDING_MAX_BYTES).toBe(4 * 1024 ** 3)
+  it('formats sizes in pt-BR without imposing a ceiling', () => {
+    expect(formatRecordingFileSize(4.6 * 1024 ** 3)).toBe('4,6 GB')
+    expect(formatRecordingFileSize(24 * 1024 ** 3)).toBe('24 GB')
   })
 
   it('formats the byte range below 1 KB honestly', () => {