Verified model record

MiniMax H3 Max Turbo

H3 Max Turbo is fal's hosted, post-trained MiniMax H3 variant, available through separate text-to-video and image-to-video endpoints. Use its endpoint-specific pricing rather than H3 Max rates. This resource has not independently benchmarked its speed or output quality.

Status
active
Developer
fal
Released
Not disclosed
Evidence
6 first-party records
Record published
Last evidence review

Identity

Model, family, and access are separate

A hosted endpoint can have different limits, timing, and pricing from its underlying model family.

Canonical name
MiniMax H3 Max Turbo
Aliases
H3 Max Turbo, fal H3 Max Turbo
Developer
fal
Available through
fal
Family relationship
Post-trained variant; see the related base-model record
Review state
Published from verified first-party claims

Specifications

Verified capabilities

MiniMax H3 Max Turbo capability record
FieldVerified valueEvidence
Input Modalitiestext, imageS4S1
Output ModalitiesvideoS4S1
Resolutions480p, 768pS4S1
Endpoint Modestext-to-video, image-to-videoS4S1

Release date, complete duration constraints, reference-input support and independent performance remain unestablished by this record. Omitted capabilities are unknown, not unsupported.

Performance

Reported speed, with its boundary

unknown

Not disclosed

No reviewed performance claim is published for this model.

Access

Published provider endpoints

Pricing

Endpoint-scoped price records

MiniMax H3 Max Turbo reviewed price versions
EndpointVariantRateStateEvidence
H3 Max Turbo Image to VideoResolution: 480p$0.00625 / secondHistorical rate — needs re-verification · checked
Review required by
S2
H3 Max Turbo Image to VideoResolution: 768p$0.01 / secondHistorical rate — needs re-verification · checked
Review required by
S2
H3 Max Turbo Text to VideoResolution: 480p$0.00625 / secondHistorical rate — needs re-verification · checked
Review required by
S5
H3 Max Turbo Text to VideoResolution: 768p$0.01 / secondHistorical rate — needs re-verification · checked
Review required by
S5

Estimate scope: model inference only. Storage, retries, moderation, transcoding, bandwidth, platform fees, taxes, and engineering are not included.

Provenance

Source ledger for this record

View source

S1 · H3 Max Turbo Image to Video API documentation

Publisher
fal
Evidence
Official documentation
Applies to
Documentation inspection only; no paid request or independent performance test. GPU inference timing is not end-to-end latency.
Checked
Open original source
View source

S2 · H3 Max Turbo Image to Video pricing

Publisher
fal
Evidence
Official documentation
Applies to
Read-only endpoint-page retrieval repeated at the documented promotion review boundary. Launch-rate wording was still visible, but the cutoff time and timezone are not stated; neither the launch rate nor the later advertised rate is published here as a current checkout price.
Checked
Open original source
View source

S3 · h3-max-turbo image-to-video official input schema

Publisher
fal
Evidence
Official documentation
Applies to
Read-only OpenAPI inspection: named endpoint POST request schema references an input duration integer with minimum 5, maximum 15, default 5. No generation request executed.
Checked
Open original source
View source

S4 · H3 Max Turbo Text to Video API documentation

Publisher
fal
Evidence
Official documentation
Applies to
Documentation inspection only; no paid request or independent performance test. GPU inference timing is not end-to-end latency.
Checked
Open original source
View source

S5 · H3 Max Turbo Text to Video pricing

Publisher
fal
Evidence
Official documentation
Applies to
Read-only endpoint-page retrieval repeated at the documented promotion review boundary. Launch-rate wording was still visible, but the cutoff time and timezone are not stated; neither the launch rate nor the later advertised rate is published here as a current checkout price.
Checked
Open original source
View source

S6 · h3-max-turbo text-to-video official input schema

Publisher
fal
Evidence
Official documentation
Applies to
Read-only OpenAPI inspection: named endpoint POST request schema references an input duration integer with minimum 5, maximum 15, default 5. No generation request executed.
Checked
Open original source

Need the complete collection? Open the site source ledger or read the methodology.