Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Mirage / Captions / Mirage Avatar

Mirage Avatar X

Generates speaking-avatar video from driving audio or a script, with reference identity controls and different native and Fal API contracts.

availableClosed weightsVideo generation
Official documentation (opens in a new tab)Reviewed 15 Oct 2026
6 of 21 specification fields collected
Output resolutionNot collectedAwaiting source review
Clip durationNative current reference: 1–180 seconds of driving audio; older guide says 1–60. Fal: 3–180 seconds from audio or up to 180 seconds from a script.See specification notes below
Input billingSee ratesMirage API
Output billingSee ratesPublished units and conditions below

Specifications

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit
Not collected

No model-specific public contract established for this field.

Maximum input tokens
Not collected

No model-specific public contract established for this field.

Maximum output
Not collected

No model-specific public contract established for this field.

Reasoning
Not collected

No model-specific public contract established for this field.

Tool calling
Not collected

No model-specific public contract established for this field.

Structured output
Not collected

No model-specific public contract established for this field.

Streaming
Not collected

No model-specific public contract established for this field.

Batch processing
Not collected

No model-specific public contract established for this field.

Fine-tuning
Not collected

No model-specific public contract established for this field.

Knowledge cutoff
Not collected

No model-specific public contract established for this field.

Release date
Not collected

No model-specific public contract established for this field.

Parameters
Not collected

No model-specific public contract established for this field.

Architecture
Not collected

No model-specific public contract established for this field.

Languages
Not collected

No model-specific public contract established for this field.

Context caching
Not collected

No model-specific public contract established for this field.

Input modalities
Audio, Image, Video, TextSource for Input modalities (opens in a new tab)

Script input is documented on the Fal route; the native route requires uploaded driving audio.

Video duration
Native current reference: 1–180 seconds of driving audio; older guide says 1–60. Fal: 3–180 seconds from audio or up to 180 seconds from a script.Source for Video duration (opens in a new tab)

See each endpoint for source scope and the native documentation conflict.

Configuration discrepancy
The October 1 native reference allows 1–180-second audio. The August 13 guide still says 1–60 seconds. Confirm account access for longer clips.Source for Configuration discrepancy (opens in a new tab)

Both versions are retained as evidence; this is not a runtime-verified extension.

Access & pricing

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

API model ID
mirage-avatar-x
Access
Documented native API; platform account and x-api-key required. No request submitted.
Regions
Not established
API endpoint URL
POST https://api.mirage.app/v1/videosSource for API endpoint URL (opens in a new tab)
Authentication
x-api-key header; multipart/form-data uploads.Source for Authentication (opens in a new tab)
Model controls
model, audio_reference, image_reference OR video_referenceSource for Model controls (opens in a new tab)
Reference guidance
Exactly one image or video reference, plus driving audio. Images: JPEG or PNG. Video: MP4 or MOV, 10–60 seconds, feature-gated preview.Source for Reference guidance (opens in a new tab)
Audio input limits
WAV or MP3; 1–180 seconds in the current schema. Earlier guide limits audio to 60 seconds.Source for Audio input limits (opens in a new tab)
Output resolutions
720pSource for Output resolutions (opens in a new tab)

Guide describes 720p MP4; current request schema exposes no resolution selection.

Configuration discrepancy
The October 1 native reference allows 1–180-second audio. The August 13 guide still says 1–60 seconds. Confirm account access for longer clips.Source for Configuration discrepancy (opens in a new tab)

Both versions are retained as evidence; this is not a runtime-verified extension.

Content provenance
Guide states outputs include non-removable provenance metadata; reference media is not retained as a reusable identity. This is not a comprehensive zero-retention promise.Source for Content provenance (opens in a new tab)
Processing workflow
Poll GET /v1/videos/{id}; download through GET /v1/videos/{id}/content after COMPLETE. Current response uses id; video_id is deprecated. Statuses: PROCESSING, COMPLETE, FAILED, CANCELLED.Source for Processing workflow (opens in a new tab)

Download redirect described by the generation guide.

Request quotas
2 generation requests per minute per organization; higher limits by account-manager arrangement. HTTP 429 when exceeded.Source for Request quotas (opens in a new tab)
Documented model selectors
mirage-avatar-x, mirage-video-1-latestSource for Documented model selectors (opens in a new tab)

Default mirage-avatar-x. The second selector is documented, but its exact checkpoint relationship to Avatar X and the 2025 research model is unresolved. Not counted as a proven separate model.

Billing basis
Round generated duration up to a multiple of six seconds; multiply by USD 0.15.Source for Billing basis (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 15 Oct 2026
CategoryUsageRateUnit
Media processingNative avatar video1$0.15output second

Rate conditions

  1. Output duration rounds up to six-second blocks: 10.2 seconds bills as 12, costing USD 1.80. Do not apply this rounding to Fal without host-specific evidence.
Pricing source (opens in a new tab)

Data & enterprise terms

API terms grant Mirage rights to use input and output to provide and improve services and develop products. This is not a no-training commitment. Exact retention period and region are not established. Model ownership remains with Mirage.

Read the service policy (opens in a new tab)
API model ID
mirage-api/avatar-x/reference-to-video
Access
Documented hosted route; Fal account and API key required. No request submitted.
Regions
Not established
API endpoint URL
https://queue.fal.run/mirage-api/avatar-x/reference-to-videoSource for API endpoint URL (opens in a new tab)
Authentication
Fal API key; SDK uses FAL_KEY environment variable.Source for Authentication (opens in a new tab)
Processing workflow
Asynchronous queue, status polling and completed MP4 download; queue status updates do not imply streaming generated media.Source for Processing workflow (opens in a new tab)
Model controls
video_reference_url, image_reference_url, avatar, audio_urlSource for Model controls (opens in a new tab)
Input modalities
Audio, Image, VideoSource for Input modalities (opens in a new tab)
Stock avatar options
View all 20 entries
  • Ayesha
  • Ayesha (16:9)
  • Farhan
  • Farhan (16:9)
  • Giulia
  • Giulia (16:9)
  • Jasmine
  • Jasmine (16:9)
  • Luke
  • Luke (16:9)
  • Maya
  • Maya (16:9)
  • Michael
  • Michael (16:9)
  • Neha
  • Neha (16:9)
  • Tariq
  • Tariq (16:9)
  • Valerie
  • Valerie (16:9)
Source for Stock avatar options (opens in a new tab)

Ten stock identities, each in portrait and landscape. These are model inputs, not separate models. None is a control value, not an additional avatar.

Reference guidance
Visual priority: video_reference_url, then image_reference_url, then avatar. Stock avatar supplies appearance only; audio_url always supplies the performance.Source for Reference guidance (opens in a new tab)
Reference audio
Required driving audio: 3–180 seconds, maximum 64 MiB.Source for Reference audio (opens in a new tab)
Audio formats
MP3, PCM WAV, AIFF, AAC, ALAC, Opus, Vorbis, FLACSource for Audio formats (opens in a new tab)

One mono or stereo audio stream, 8–48 kHz.

Input video limits
Reference video: 1–120 seconds, maximum 1 GiB. H.264, H.265/HEVC or ProRes in MP4/MOV/M4V; VP8/VP9 in WebM. Embedded audio is ignored.Source for Input video limits (opens in a new tab)

Exactly one video stream and a usable one-second reference window required.

Aspect ratios
9:16, 16:9Source for Aspect ratios (opens in a new tab)

Reference display aspect ratio must be within 5% of the target.

Reference media access
Public HTTPS media URLs without cookies or authorization headers. Model-specific requirements take precedence over generic SDK data-URI examples.Source for Reference media access (opens in a new tab)
Video duration
Follows driving audio, 3–180 seconds.Source for Video duration (opens in a new tab)
Billing basis
USD 0.30 per generated-video second. Subsecond rounding is not specified in the reviewed Fal page.Source for Billing basis (opens in a new tab)
Image input limits
One static JPG/JPEG/PNG/WebP, maximum 25 MiB; short edge at least 512 px, no edge over 8192 px, at most 25 megapixels.Source for Image input limits (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 15 Oct 2026
CategoryUsageRateUnit
Media processingFal audio-driven avatar video1$0.30output second

Rate conditions

  1. Host-specific generated-video duration charge. Native six-second rounding is not established for Fal. No text-question estimate or token tariff applies.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

API model ID
mirage-api/avatar-x/text-to-video
Access
Documented hosted route; Fal account and API key required. No request submitted.
Regions
Not established
API endpoint URL
https://queue.fal.run/mirage-api/avatar-x/text-to-videoSource for API endpoint URL (opens in a new tab)
Authentication
Fal API key; SDK uses FAL_KEY environment variable.Source for Authentication (opens in a new tab)
Processing workflow
Asynchronous queue, status polling and completed MP4 download; queue status updates do not imply streaming generated media.Source for Processing workflow (opens in a new tab)
Model controls
video_reference_url, avatar, audio_reference_url, scriptSource for Model controls (opens in a new tab)
Input modalities
Text, Audio, VideoSource for Input modalities (opens in a new tab)
Stock avatar options
View all 20 entries
  • Ayesha
  • Ayesha (16:9)
  • Farhan
  • Farhan (16:9)
  • Giulia
  • Giulia (16:9)
  • Jasmine
  • Jasmine (16:9)
  • Luke
  • Luke (16:9)
  • Maya
  • Maya (16:9)
  • Michael
  • Michael (16:9)
  • Neha
  • Neha (16:9)
  • Tariq
  • Tariq (16:9)
  • Valerie
  • Valerie (16:9)
Source for Stock avatar options (opens in a new tab)

Ten stock identities, each in portrait and landscape. These are model inputs, not separate models. None is a control value, not an additional avatar.

Reference guidance
Default avatar Jasmine. Custom video_reference_url and audio_reference_url must be provided together and replace the stock visual and voice.Source for Reference guidance (opens in a new tab)
Reference audio
Optional voice reference paired with a video: 10–180 seconds, maximum 64 MiB; only the first 60 seconds are used.Source for Reference audio (opens in a new tab)
Audio formats
MP3, PCM WAV, AIFF, AAC, ALAC, Opus, Vorbis, FLACSource for Audio formats (opens in a new tab)

One mono or stereo audio stream, 8–48 kHz.

Input video limits
Reference video: 1–120 seconds, maximum 1 GiB. H.264, H.265/HEVC or ProRes in MP4/MOV/M4V; VP8/VP9 in WebM. Embedded audio is ignored.Source for Input video limits (opens in a new tab)
Aspect ratios
9:16, 16:9Source for Aspect ratios (opens in a new tab)

Reference display aspect ratio must be within 5% of the target.

Reference media access
Public HTTPS media URLs without cookies or authorization headers. Model-specific requirements take precedence over generic SDK data-URI examples.Source for Reference media access (opens in a new tab)
Video duration
Generated speech at most 180 seconds; speaking pace can cause even a short script to exceed this.Source for Video duration (opens in a new tab)
Billing basis
USD 0.30 per generated-video second. Subsecond rounding is not specified in the reviewed Fal page.Source for Billing basis (opens in a new tab)
Prompt length limit
Script must contain 50–1500 characters. The minimum typically produces about four seconds, not a guaranteed duration.Source for Prompt length limit (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 15 Oct 2026
CategoryUsageRateUnit
Media processingFal script-driven avatar video1$0.30output second

Rate conditions

  1. Host-specific generated-video duration charge. Native six-second rounding is not established for Fal. No text-question estimate or token tariff applies.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

Caching

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

Mirage API

mirage-avatar-x
Not verified

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

How it works
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Minimum reusable input
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Lifetime & refresh
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

How to enable it
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Usage fields
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Billing rules
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Limits & data boundaries
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Caching reviewed 15 Oct 2026

fal.ai

mirage-api/avatar-x/reference-to-video
Not verified

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

How it works
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Minimum reusable input
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Lifetime & refresh
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

How to enable it
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Usage fields
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Billing rules
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Limits & data boundaries
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Caching reviewed 15 Oct 2026

fal.ai

mirage-api/avatar-x/text-to-video
Not verified

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

How it works
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Minimum reusable input
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Lifetime & refresh
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

How to enable it
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Usage fields
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Billing rules
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

Limits & data boundaries
Not collected

No inference-cache contract or discount established. A reusable avatar, stored voice, job ID or legal permission to cache content does not establish prompt caching.

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Caching reviewed 15 Oct 2026

What the measurements say

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

The record behind the model

Why it is included

First-party attribution identifies a Mirage-developed model. Closed-weight status is inferred from its hosted distribution and proprietary-model ownership terms; the reviewed sources establish no downloadable checkpoint or open-weight license.

Classification evidence (opens in a new tab)

Sources reviewed