Primary use stated or inferred cautiously from official documentation.
N
NVIDIA / nvidia/Metatron-v1
Metatron v1
A control policy for Unitree G1 with 29 actions: locomotion, recovery and scene interaction. ONNX models appeared on October 27; the native G1 package was published on the 28th.
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/Metatron-v1 |
|---|---|
| Model type | Robotics |
| Access model | Open weights |
| License | NVIDIA Open Model License |
| Deployment | Self-hosted |
| Release | 27 OCT 2026 |
| Knowledge cutoff | Not published |
| Entrance | Robot state |
| Exit | Robot actions |
| Context window | Motion history and goals |
| maximum output | 29 actions |
| Reasoning | Not published |
| Published tools | |
| Structured Outings | Not published |
| Batch processing | Not published |
| Prompt cache | Not published |
| Fine-tuning | Not published |
| Verified platforms | PyTorch · ONNX · ProtoMotions |
Control and recovery research in compatible simulators. Native weights and browser exports retain different input contracts.
The native preset disables scenes even though the weights support them. It is not interchangeable with the 39-input ONNX. Hardware use requires supervision, limits and an emergency stop.
03 / CAPABILITIES
What it can do
Orientation
Control and recovery research in compatible simulators. Native weights and browser exports retain different input contracts.
Context
Motion history and goals · 29 actions
Tools and integration
Not published
Access
Hugging Face · downloadable checkpoint
Input modalities
Robot state
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 |
Model under the NVIDIA Open Model License; ProtoMotions code under Apache 2.0, with separate data and asset rights. There is no verified per-token rate; simulation, hardware and operations have their own costs.
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 card documents integration and export checks without a standardized robot-success benchmark. Technical compatibility is not safety certification or independent reproduction.
i No verified scores for this checkpoint.
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
Metatron v1
Version added to Inferama's verified catalog.
07 / SOURCES