Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Meta / Emu

Emu

Meta’s aesthetic text-to-image model using a large latent-diffusion U-Net and a small, curated quality-tuning dataset.

Image generation
restrictedClosed weightsImage generation
Official documentation (opens in a new tab)Reviewed 13 Oct 2026
7 of 18 specification fields collected
Output resolution1024 × 1024See specification notes below
Output formatsNot collectedAwaiting source review
Input billingNot collectedMeta · research reference
Output billingNot collectedNo inferred price

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
Maximum output
Not collected
Input modalities
TextSource for Input modalities (opens in a new tab)
Output modalities
ImageSource for Output modalities (opens in a new tab)
Reasoning
Not collected
Tool calling
Not collected
Structured output
Not collected
Streaming
Not collected
Batch processing
Not collected
Fine-tuning
Not collected
Knowledge cutoff
Not collected
Release date
2023-09-27Source for Release date (opens in a new tab)

Research disclosure date, not general API or weight availability.

Parameters
2.8B trainable U-Net parameters; excludes encoders and autoencoderSource for Parameters (opens in a new tab)
Architecture
Latent diffusion, 16-channel autoencoder, CLIP ViT-L and T5-XXL text encodersSource for Architecture (opens in a new tab)
Languages
Not collected
Context caching
Not collected

No model-specific inference service or discounted prompt-cache contract is established for these research weights.

Output resolutions
1024 × 1024Source for Output resolutions (opens in a new tab)
Published training data
1.1 billion image-text pairs; 2,000 curated images for quality tuningSource for Published training data (opens in a new tab)

Access & pricing

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

Meta · research reference

Provider source (opens in a new tab)
API model ID
Not collected

Descriptive catalog identity, not a verified selectable API ID.

Access
Paper and examples available. No original public checkpoint or callable inference contract verified.
Regions
Research deployment only; public availability and residency are not established

Endpoint pricing has not been collected. Check the provider for applicable rates.

Data & enterprise terms

The cited research release does not supply a current production processing/retention agreement. Research access or an expired demonstration is not a production API entitlement.

Read the service policy (opens in a new tab)

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.

Meta · research reference

Not verified

No published service cache policy for this research record.

How it works
Not collected

No published service cache policy for this research record.

Minimum reusable input
Not collected

No published service cache policy for this research record.

Lifetime & refresh
Not collected

No published service cache policy for this research record.

How to enable it
Not collected

No published service cache policy for this research record.

Usage fields
Not collected

No published service cache policy for this research record.

Billing rules
Not collected

No published service cache policy for this research record.

Limits & data boundaries
Not collected

No published service cache policy for this research record.

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

What the measurements say

Visual preference versus pretraining-only counterpart

Meta research authors · provider-reported
82.9% preference

Provider human visual-appeal comparison against its own pretraining-only baseline, not general task accuracy.

Evaluation source (opens in a new tab)

PartiPrompts visual preference versus SDXL 1.0

Meta research authors · provider-reported
68.4% preference

Provider-reported visual-appeal comparison on PartiPrompts at the 2023 release. Not a current ranking.

Evaluation source (opens in a new tab)

Open User Input visual preference versus SDXL 1.0

Meta research authors · provider-reported
71.3% preference

Provider-collected real-user-prompt evaluation, distinct from PartiPrompts.

Evaluation source (opens in a new tab)

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

Original Meta research weights are not supplied by the reviewed release. The paper documents this named model or explicitly identified trained variant, while public artifacts provide research descriptions, examples or evaluation data rather than an original downloadable checkpoint. This record covers those unreleased research weights, not independently trained reimplementations or openly available base components.

Classification evidence (opens in a new tab)

Lifecycle dates

Research model disclosed27 Sept 2023Source (opens in a new tab)

Sources reviewed