The Nox model family

Nox runs on a small family of purpose-built models. Each entry below states what the model is — its architecture, context window, reasoning behavior, modalities, and published measurements — and the plan it ships on.

Models

Nox 1.3 Astra (Flagship)

The flagship — deepest reasoning in the family

1.3 Astra is Nox's flagship model. It is a 700-billion-parameter mixture-of-experts reasoning engine that activates 36.9 billion parameters per token, reads a context window of roughly one million tokens, and runs a deliberate reasoning pass before every answer.

Available on MAX.

More about Nox 1.3 Astra

Nox 1.3 Nova

The balance of depth and speed

1.3 Nova sits directly below Astra in the Nox 1.3 family. It is a hybrid-reasoning model: it decides per question whether to run a step-by-step reasoning pass or answer directly, which keeps replies conversational in speed while staying structured in substance.

Available on Pro and MAX.

More about Nox 1.3 Nova

Nox 1.3 Vela

The fast 1.3 model on every plan

1.3 Vela is the everyday model of the 1.3 family and the one available on every plan, including Free. It runs on a lean 21-billion-parameter mixture-of-experts engine that activates only about 3.6 billion parameters per token — a small active set that is exactly what makes it near-instant.

Available on every plan, including Free.

More about Nox 1.3 Vela

Nox Light 1.1

The previous-generation fast model — retiring soon

Light 1.1 is the previous-generation model behind everyday Nox conversations, tuned for low latency and concise answers. It is being retired: Nox 1.3 Vela is its successor on every plan, and Light stays available under “More models” during the transition.

Legacy — retiring. Available on every plan under “More models” during the transition; Nox 1.3 Vela is its successor.

More about Nox Light 1.1

Nox Agent 1.2

Deep research with live web sources

Nox Agent 1.2 is the research model of the family. Instead of producing a single reply, it plans a question out, gathers material across multiple steps, and returns a direct answer together with resource cards — each carrying a title, its source, and a link — so the answer can be traced instead of taken on trust.

Available on Pro and MAX.

More about Nox Agent 1.2

Nox Voice 1.2 (Beta)

Talk to Nox in real time

Nox Voice 1.2 (beta) is the family's real-time speech model, built on leading conversational voice AI technology. It runs a live, low-latency speech loop: spoken input in, natural spoken answers out, with the full exchange transcribed into the chat thread as it happens.

Available on MAX. Currently in beta.

More about Nox Voice 1.2