neurok@studio:~$ neurok start --desktop=nova
[ 0.001s] engine ............ OK resonance-core 5.1 [ 0.214s] language ......... OK en / zh-CN [ 0.409s] research ......... OK 7 programmes loaded [ 0.512s] safety ........... OK scaling policy v3 active [ 0.688s] models ........... OK cortex / pulse / echo [ 0.741s] desktop .......... READY
The same site,
as a command surface.
Everything the studio desktop can do — research, safety, models, news, company, learn — re-expressed as a terminal. Read it like a log. Drive it like a console.
01 // COMMANDS
Point anywhere
researchSeven programmes — interpretability, alignment, engineering, security, economics, science, impact.
safetyScaling thresholds T1–T4, the model constitution, transparency, and the public incident log.
modelsCortex for long-horizon work, Pulse as the default, Echo for volume. One API surface.
newsReleases, papers, and corrections — one feed, filterable.
companyStructure, leadership, open roles, and how to reach a human.
learnAcademy courses, runnable tutorials, docs versioned with the models.
02 // PROCESSES
Models, running
PIDMODELROLECTXCOSTSTATUS
0x01cortex-5.1long-horizon agents1M$$$$online
0x02pulse-5.1default / production1M$$online
0x03echo-5routing / volume400K$online
Swap the model name; the rest of the code keeps working. see api →
03 // TAIL -f NEWS
Latest output
2026-09-01release: cortex-5.1 + pulse-5.1 — one benchmark figure withdrawn, see system card
2026-07-24paper: a steerable circuit for answer withholding — three failed replications included
2026-06-30policy: responsible scaling v3 — thresholds now carry named safeguards
2026-06-02index: task-level exposure opened to external review