Genesis AIFrance2026
GENE-26.5
An AI model: a vision-language-action model (VLA) takes camera images and an instruction and turns them into movements for a robot. This page describes the software, not a robot.
Robotsnot stated
Key figures as stated by the developer
2 of 6 stated · 4 gaps- Model type not stated
-
Controls A
arms and hands -
Inputs A
cameralanguagerobot statetouch - Access not stated
- Parameters not stated
- Base model not stated
How to read the figures
- 14kgA30.9lbA
- The letter in the grey square after a figure points to the source the figure came from. Robots with several sources carry several letters.
- 74.5lb
- Imperial figures are our own conversion unless the manufacturer states an imperial figure itself. The source letter always belongs to the metric figure — a conversion has no source of its own.
Versions and notes in the key figures (4)
Model type
Note Genesis AI uses none of the three types for the model: "our first robotic foundation model system and the initial public release in the GENE family". It names vision-language models only as one source of priors: "Vision-Language models (VLMs) encode intent and semantic representations".
Controls
"GENE-26.5 currently runs on a hardware platform that is already highly dexterous"; "we replaced the vendor-supplied controller on our bimanual robotic arms". The blog shows a two-armed setup with hands and does not list the outputs as such.
Inputs
Genesis AI builds "a unified model that best absorbs scale across heterogeneous inputs and outputs": language, vision, proprioception, tactile and action. Vision = camera, proprioception = robot state, tactile = touch.
Base model
Note The model draws on priors from pre-trained models: "Vision-Language models (VLMs) encode intent and semantic representations".
Full specifications
3 of 12 fields stated| Field | Value | Source and caveats |
|---|---|---|
| Input and output | ||
| Model type | not stated | Note Genesis AI uses none of the three types for the model: "our first robotic foundation model system and the initial public release in the GENE family". It names vision-language models only as one source of priors: "Vision-Language models (VLMs) encode intent and semantic representations". A |
| Inputs | camera language robot state touch | A Genesis AI builds "a unified model that best absorbs scale across heterogeneous inputs and outputs": language, vision, proprioception, tactile and action. Vision = camera, proprioception = robot state, tactile = touch. |
| Controls | arms and hands | A "GENE-26.5 currently runs on a hardware platform that is already highly dexterous"; "we replaced the vendor-supplied controller on our bimanual robotic arms". The blog shows a two-armed setup with hands and does not list the outputs as such. |
| Degrees of freedom controlled | not stated | Note The only count is for the hand, which is a later platform, not the one the model runs on: "It features 20 active, back-drivable degrees of freedom"; "Genesis Hand 1.0 represents the next step in our hardware roadmap". A |
| Size and speed | ||
| Parameters | not stated | |
| Control rate | not stated | Note Control middleware for the arms, not the model: the system "uses KickCAT as the EtherCAT master with Distributed Clocks support, runs at 500 Hz". A |
| Where it runs and what it is built on | ||
| Runs where | not stated | |
| Base model | not stated | Note The model draws on priors from pre-trained models: "Vision-Language models (VLMs) encode intent and semantic representations". A |
| Openness | ||
| Access | not stated | |
| License | not stated | |
| The developer’s own claims | ||
| Learns new tasks | needs retraining | A Developer claim, under task-specific fine-tuning: "We curate novel tasks fully excluded from pre-training and evaluate them in a ultra-low-data regime", "each with ~20–30 minutes of data". |
| Works across robots | not stated | Note The blog describes one hardware platform: "GENE-26.5 currently runs on a hardware platform that is already highly dexterous". A Yes only if the developer itself says the model runs on different robot models, for example robots from two manufacturers or two of one manufacturer’s models. Several units of the same robot do not count. |
Notes
- Genesis AI calls GENE-26.5 an AI foundation model, purpose-built for robotics, and the first release in the GENE family.
- Genesis AI announced GENE-26.5 on 6 May 2026 together with a dexterous robot hand and a data engine.
Sources
-
A
Manufacturer product page. www.
genesis. · Retrievedai/ blog/ gene-26-5-advancing-robotic-manipulation-to-human-level -
B
www.
genesis. · Retrievedai/ press/ press-release-gene-265
One line per source, not per figure: a source has one retrieval date. The letter beside a figure points to the line it came from.