Changelog | Uthana API Docs

On this page

All notable changes to the Uthana API will be documented in this file.

July 9, 2026

Changed

text-to-motion-3.0 is now available to any account on the pay-as-you-go plan. Use the new create_text_to_motion_job mutation with model: "text-to-motion-3.0". See Pricing for details.P

Added

text-to-motion-3.0 is now available in limited release via the new create_text_to_motion_job mutation. This async, diffusion-based model returns a Job that must be polled to completion, similar to video-to-motion. Contact support to request access.

Changed

video-to-motion-2.0 is now the default VTM model (previously video-to-motion-v2). video-to-motion-v2 remains accepted as a backward-compatible alias.
Added video-to-motion-2.1 model with post-processing refinements. Returns refined motion IDs rather than the base motion.

Removed

video-to-motion-v1 is no longer available. The deprecation deadline of 30 Apr 2026 has passed.

Added

playback_speed download option — scale animation playback speed between 0.1–2.0x (default 1.0).
rerig_target and include_fingers options for character upload, for re-rigging to a specific target skeleton and including finger joints. Available in @uthana/client v0.3.1 and uthana (Python) v0.4.0.
Client SDK usage examples added throughout the API docs (Python and JS/TS tabs).

Added

@uthana/client and @uthana/react JS/TS client packages published on npm — v0.2.0 initial release.
Python uthana package v0.2.0 — expanded API covering all capabilities with both async and sync variants.
Locomotion API: create_locomotion mutation for controllable, loopable directional travel motions. JS/Python clients updated to v0.3.0 with locomotion support.
no_mesh option for GLB downloads — returns animation data only, without the character mesh.

Changed

video-to-motion-v2 is now the default VTM model. video-to-motion-v1 is deprecated and will be removed.

Added

Stitch & loop motions: new create_enhanced_stitched_motion mutation transitions smoothly between two existing motions. See Stitch & loop motions.
BUCMD diffusion-v2 model for create_text_to_motion with advanced parameters: steps, cfg_scale, length, and seed.
auto_rig_confidence field in character upload responses — a 0–1 quality score for the auto-rigging result.
in_place, torso_only, and roblox_compatible download options for FBX/GLB exports.

Added

Motion-only animation download endpoint: /motion/animation/{character_id}/{motion_id}/glb/... returns just the animation track without a character mesh, and does not count against download quota.

Added

Changed