Sereact · robot AI model
Cortex
Germany · 2025
Sources retrieved: 30 October 2026 (A, B, C, D, E, F)
- Model type
- Vision-language-action (VLA)Source and note: Model typeC sereact.ai/posts/teradyne-robotics-partnership · Retrieved 30 October 2026Note. Developer wording: “Cortex, Sereact's Vision-Language-Action (VLA) model”. Cortex 2.0 keeps it: “Rather than replacing the existing VLA, the world model complements it”.
- What it controls
- Arms and handsSource and note: What it controlsA sereact.ai/posts/cortex-2 · Retrieved 30 October 2026Note. Cortex 2.0 post: the VLA will “emit action chunks that route into embodiment-specific low-level controllers”. The robots Sereact names are arms: Cortex “runs on two Universal Robots UR12e-1300s”.
- Inputs
- Camera Depth Robot state LanguageSource and note: InputsA sereact.ai/posts/cortex-2 · Retrieved 30 October 2026Note. Cortex 1.0 architecture, as described in the Cortex 2.0 post: it “turns multi-modal perception, RGB, depth/3D geometry, proprioception, and task context, into a control signal”. Its task representation is built to “align language with scene entities and constraints”.
Specifications
6/14 specifications disclosed 8 not stated
7 fields hidden 1 not-stated field with a note stays ·
How to read the figuresShow guideClose
Not statedThe developer publishes nothing on itYesThe developer says it does
Select a source marker for source details and notes.
Release1/1 disclosed
Released29 September 2025
Source and note: ReleasedF tech.eu/2025/09/29/meet-cortex-sereact-s-foundation-model-that-lets-robots-think-on-their-feet/ · Retrieved 30 October 2026Note. Tech.eu article on the first Cortex: “29 September 2025”.
Input and output3/4 disclosed
Model typeVision-language-action (VLA)
Source and note: Model typeC sereact.ai/posts/teradyne-robotics-partnership · Retrieved 30 October 2026Note. Developer wording: “Cortex, Sereact's Vision-Language-Action (VLA) model”. Cortex 2.0 keeps it: “Rather than replacing the existing VLA, the world model complements it”.
InputsCamera Depth Robot state Language
Source and note: InputsA sereact.ai/posts/cortex-2 · Retrieved 30 October 2026Note. Cortex 1.0 architecture, as described in the Cortex 2.0 post: it “turns multi-modal perception, RGB, depth/3D geometry, proprioception, and task context, into a control signal”. Its task representation is built to “align language with scene entities and constraints”.
What it controlsArms and hands
Source and note: What it controlsA sereact.ai/posts/cortex-2 · Retrieved 30 October 2026Note. Cortex 2.0 post: the VLA will “emit action chunks that route into embodiment-specific low-level controllers”. The robots Sereact names are arms: Cortex “runs on two Universal Robots UR12e-1300s”.
Degrees of freedom controlledNot stated
Size and speed0/2 disclosed
ParametersNot stated
Control rateNot stated
Where it runs and what it is built on0/2 disclosed
On robot or cloudNot stated
Base modelNot stated
Source and note: Base modelD sereact.ai/posts/cortex-2-learns-from-practice · Retrieved 30 October 2026Note. Cortex 2.6 post, named without a model name: “the VLA backbone of our Cortex 2.6, pretrained on industrial robot data”.
Openness0/3 disclosed
AccessNot stated
Licence documentNot stated
Licence typeNot stated
The developer’s own claims2/2 disclosed
Learns new tasksFrom a few examples
Source and note: Learns new tasksE sereact.ai/posts/cortex-2-5 · Retrieved 30 October 2026Note. Developer claim for Cortex 2.5: it can “show the robot a handful of video demonstrations of a new task, and have it carry the task out, no retraining in between”. Limit stated: “tested only on tasks that are similar to ones Cortex already knows”.
Works across robotsYes
Source and note: Works across robotsB sereact.ai/product/humanoid · Retrieved 30 October 2026Note. Developer claim: “Cortex runs the same intelligence on different robots. You choose the body.” Named robots: “a bimanual manipulation platform with dual UR5e arms” (Cortex 2.0 post) and “runs on two Universal Robots UR12e-1300s” (Teradyne post).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.
Versions
- Cortex 1.0 29 September 2025
- Cortex 1.5 30 January 2026
- Cortex 1.6 10 February 2026
- Cortex 2.0 21 February 2026
- Cortex 2.5 15 July 2026
- Cortex 2.6 15 October 2026
The developer has released this model in several versions. Each figure applies to the version its source names.
Notes
Sereact publishes Cortex as numbered versions; the newest version on its blog is Cortex 2.6 (15 October 2026).
Tech.eu presented Cortex on 29 September 2025 as a Vision-Language-Action model.
Cortex 2.0 (21 February 2026) adds a world model that predicts and scores future outcomes before the robot acts.
Sereact GmbH is registered at an address in Stuttgart (legal notice on sereact.ai).
Sources
One line per source, not per figure: a source has one retrieval date.
- Asereact.ai/posts/cortex-2Main source. Retrieved 30 October 2026
- Bsereact.ai/product/humanoidRetrieved 30 October 2026
- Csereact.ai/posts/teradyne-robotics-partnershipRetrieved 30 October 2026
- Dsereact.ai/posts/cortex-2-learns-from-practiceRetrieved 30 October 2026
- Esereact.ai/posts/cortex-2-5Retrieved 30 October 2026
- Ftech.eu/2025/09/29/meet-cortex-sereact-s-foundation-model-that-lets-robots-think-on-their-feet/Secondary. Retrieved 30 October 2026