Tripo Texture 3.0

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 input tokens

Not collected

Maximum output

Not collected

Reasoning

Not collected

Tool calling

Not collected

Structured output

Not collected

Task JSON is not user-defined structured model output.

Streaming

Not collected

Polling and completion webhooks do not establish streamed geometry.

Batch processing

Not collected

POST /tasks/list batches status queries, not model generation or a discounted Batch API.

Fine-tuning

Not collected

Customer weight fine-tuning is not documented.

Knowledge cutoff

Not collected

Release date

Not collected

Parameters

Not collected

Architecture

Not collected

Languages

Not collected

Context caching

Not collected

Prompt caching is not documented.

Authentication

Bearer API key; JSON request body.Source for Authentication (opens in a new tab)

Request quotas

Per-key request limits vary by endpoint. Account-wide 5 concurrent tasks shared within the model processing category. HTTP 429: code 1007 for request limits, 2000 for concurrency.Source for Request quotas (opens in a new tab)

These are category pools, not separate allowances per model. No fixed requests-per-second quota is published here.

Asynchronous generation

SupportedSource for Asynchronous generation (opens in a new tab)

Task ID followed by status polling or webhook; not progressive model-output streaming.

Task refunds

Credits are frozen when a task is created, consumed on success, and released on failure or cancellation.Source for Task refunds (opens in a new tab)

Data retention

Not collected

Task expiry does not establish a retention duration.

Input modalities

3D mesh, Text, ImageSource for Input modalities (opens in a new tab)

Output modalities

Textured 3D meshSource for Output modalities (opens in a new tab)

Usage metering

Native USD task prices. Texture quality and optional processing can add charges; not billed by text tokens.Source for Usage metering (opens in a new tab)

PBR materials

SupportedSource for PBR materials (opens in a new tab)

PBR enabled by default; existing geometry is not regenerated.

Seed control

texture_seed controls generated texture.Source for Seed control (opens in a new tab)

Output transport

Asynchronous task result containing model and preview URLs.Source for Output transport (opens in a new tab)

Output formats

GLBSource for Output formats (opens in a new tab)

GLB is shown in the result example; this is not an exhaustive export-format guarantee. H-series quad generation explicitly forces FBX.

Input file formats

GLB, GLTF, FBX, OBJ, STLSource for Input file formats (opens in a new tab)

Maximum input content size

150,000,000Source for Maximum input content size (opens in a new tab)

150 MB mesh upload; decimal byte conversion.

Standalone texturing

SupportedSource for Standalone texturing (opens in a new tab)

Required request fields

One mesh source: URL, file_token or a prior generation task_id. Text, one image, or exactly four reference images are mutually exclusive guidance modes; style_image may accompany text.Source for Required request fields (opens in a new tab)

Reference images

Four-image guidance order: front, left, back, right. Recommended reference resolution at least 256 × 256 pixels.Source for Reference images (opens in a new tab)

Texture resolution

standard, detailed (HD), extreme (8K). Fast is rejected.Source for Texture resolution (opens in a new tab)

Mesh processing controls

bake=true by default; optional part_names limits texturing to segmented parts. delight is ignored on this version. Optional geometry compression uses meshopt.Source for Mesh processing controls (opens in a new tab)

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

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

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.

Why it is included

Qualified inference from provider-hosted named models and terms restricting recovery of underlying models. Reviewed public VAST repositories publish separate TripoSR, TripoSG and TripoSF weights, not these named H/P/Texture API selections. Internal checkpoint identities and individual weight licenses are not disclosed; this is not exhaustive proof of absence.

Sources reviewed