Primary use stated or inferred cautiously from official documentation.
N
NVIDIA / nvidia/SOMA-X · v0.3.5
SOMA-X v0.3.5
A parametric human-body system for reconstruction, animation and simulation, with an optional neural corrective model. Version 0.3.5, published on November 2, updates the corrective artifact for uncommon poses. Its geometric core is analytical: it is neither a video generator nor a language model.
01 / SUMMARY
The essential
Maximum input capacity when published by the source.
Documented maximum generation limit.
Verified availability channels.
Primary functional family and verified specialties.
Declared terms for API access, model weights, or self-hosting.
02 / TECHNICAL SHEET
Limits and integration
| API ID | nvidia/SOMA-X · v0.3.5 |
|---|---|
| Model type | 3D and motion · Robotics |
| Access model | Open weights |
| License | Apache 2.0 |
| Deployment | Self-hosted |
| Release | 02 NOV 2026 |
| Knowledge cutoff | Not published |
| Entrance | Pose and identity tensors |
| Exit | 3D meshes and joints |
| Context window | Pose and identity parameters |
| maximum output | Body and hand geometry |
| Reasoning | Not published |
| Published tools | |
| Structured Outings | Not published |
| Batch processing | Not published |
| Prompt cache | Not published |
| Fine-tuning | Not published |
| Verified platforms | Hugging Face · PyTorch · NVIDIA Warp |
Research avatars, body reconstruction and simulation with explicit pose and identity control.
The core has no learned neural parameters; the optional corrective model does. External identity models may require separate licenses. Speed figures from earlier versions do not establish 0.3.5 performance.
03 / CAPABILITIES
What it can do
Orientation
Research avatars, body reconstruction and simulation with explicit pose and identity control.
Context
Pose and identity parameters · Body and hand geometry
Tools and integration
Not published
Access
Hugging Face and GitHub
Input modalities
Pose and identity tensors
04 / PRICES
Documented cost
| Concept | Worth | Unit/condition |
|---|---|---|
| Standard input | No published official API rate | Standard API rate |
| Cached input | Not published | Reading reused prefixes |
| Cache write or storage | Not published | The condition varies by provider |
| Standard output | Self-hosted infrastructure cost | May include reasoning tokens |
| Batch input | Not published | Asynchronous processing |
| Batch output | Not published | Asynchronous processing |
The project declares Apache 2.0, with separate terms for external resources such as MANO or SMPL. No commercial API or per-token pricing is verified. Budget for infrastructure and review each backend’s licenses.
i Prices change and may depend on level, region or context length. Check the source before making a decision.
05 / EVALUATIONS
How to read the results
“
The release note describes an improvement for uncommon poses, without a specific quantitative score. Overall measurements from previous versions are not attributed to the new corrective model. Evaluation should measure geometric errors and pose stability on your own data.
i Documented qualitative change; no verified benchmark specific to 0.3.5.
Inferama only highlights a “best result” when the metric, test set, configuration, and date allow for an equivalent comparison. The supplier's figures are presented as claims from its own source.
06 / VERSIONS
Chronology
SOMA-X v0.3.5
Version added to Inferama's verified catalog.
07 / SOURCES