The Nox model family
The Nox 2 generation launched on September 23, 2026: Orion 2 upgrades the flagship, Orion Flash 2 is the new fast Orion, and Vela 2 is the everyday model. 1.3 Astra was retired on October 5, 2026.
Nox 2.5 leads the current lineup: Orion 2.5, the new flagship, and Orion Flash 2.5, the fast Orion, now publicly available in Plan mode on Pro and MAX, with workspace tools built in, health research tools from High thinking up, a staged review at Ultra, and any topic in Plan mode. HealthBenchAI lists Orion 2.5 Ultra at 69.5 length-adjusted on HealthBench Professional; Orion Flash 2.5 has not been benchmarked yet. The Nox 2.1 lineup stays available. Vela 2.5 is the everyday model on every plan and reads images. Nox Voice 2 brings spoken conversation to every paid plan, while the catalogue preserves retired models with their retirement dates and named successors.
Models
Nox GHI 1 (Coming November 1)
The officially released model system that powers Pebs, Nox's agent
Nox GHI 1 is officially released. It is Nox's in-house model system powering Pebs, the personal AI agent experience. Given a goal, it writes a step plan, executes each step with tools — live web, an isolated cloud browser, files, Nox Memories — and returns a user-facing result per step. Routine browser steps run on a lightweight engine and escalate automatically to a stronger one when a step stops making progress; the thinking level you choose sets the engine for the harder steps.
Launching November 1, 2026 on MAX and NOX 50. Each model call is metered against your plan's credits.
Nox Orion 2.5 (Flagship)
The new flagship — workspace tools built in, health research tools, and a staged review at Ultra
Orion 2.5 is the flagship of the Nox 2.5 generation, now publicly available in Plan mode on Pro and MAX. It runs the same engine as Orion 2.1 — an 824-billion-parameter mixture-of-experts model, 46 billion active per token, with a roughly one-million-token context window — and the change is in what it does around the answer: the workspace tools are built in, health questions get dedicated research tools from High thinking up and a staged review at Ultra, and in Plan mode it answers any topic. There is no separate Pro version; the tools are part of the model.
Now publicly available in Plan mode on Pro and MAX.
Nox Orion Flash 2.5 (New)
The fast Orion 2.5 — native photo input, workspace tools built in, and the same review at Ultra
Orion Flash 2.5 is the fast model of the Nox 2.5 generation, now publicly available in Plan mode on Pro and MAX. It runs the same engine as Orion Flash 2.1 — a 552-billion-parameter mixture-of-experts model, 8 billion parameters active per token during prefill and 16 billion during decode, with native photo input and a roughly one-million-token context window — and gains the tools of Orion 2.5: the workspace, health research tools from High thinking up, a staged review at Ultra, and any topic in Plan mode, while answering faster and consuming less included usage than Orion 2.5.
Now publicly available in Plan mode on Pro and MAX.
More about Nox Orion Flash 2.5
Nox Orion 2.6 Cyber Preview (Preview)
A dedicated security model for defensive and authorized security work — hardened against misuse and prompt injection
Orion 2.6 Cyber Preview focuses on one thing: security. It brings the depth of the Orion 2 generation's engine to vulnerability analysis, secure code review, detection and response, cryptography review, cloud and container security, and penetration-testing methodology — while being built to stay on the right side of the line. Access is limited, and the model is governed by a security contract that requires authorization for anything intrusive, refuses malware and real-world attack tooling by the nature of the request, and is hardened against being talked or tricked into causing harm. It is not an everyday-chat model, and not a tool for attacking systems you do not own.
Limited preview for authorized security use, in Plan mode.
More about Nox Orion 2.6 Cyber Preview
Nox Orion 2.1
The Nox 2.1 flagship, with a peer-review answer pipeline
Orion 2.1 is the flagship of the Nox 2 generation and the successor to Orion 2. The engine is unchanged — an 824-billion-parameter mixture-of-experts model, 46 billion active per token, with a roughly one-million-token context window — and the change is in how the answer is produced: a peer-review pipeline that briefs the case, writes a specialist checklist before drafting, drafts against it, and has the draft reviewed before anything reaches you. It was designed from the point losses measured on HealthBench Professional: missing named specifics, embedded errors in pasted plans that went unflagged, decisive context that was never asked for, mixed-language replies, and over-long answers.
Available in Plan mode on Pro and MAX; selectable in NOX WORK.
Nox Orion 2.1 Pro (New)
Visualize your body: 3D models, code, and MP4 video
Orion 2.1 Pro is Orion 2.1 with a workspace. The model is unchanged — an 824-billion-parameter mixture-of-experts engine with the Peer Review answer pipeline — and it now has tools: it builds interactive pages, charts, calculators and complex 3D models, checks that they run in an isolated browser sandbox, and renders MP4 video entirely from code. It is built for health: visualize your heart, spine or lungs in 3D, turn a lab report into a chart, or make a short explainer video for a patient or a class.
Available in Plan mode on Pro and MAX, on the web app.
Nox Orion Flash 2.1 (New)
The fast Orion on a new engine — 552B mixture-of-experts, native photo input
Orion Flash 2.1 is the fast version of Orion and replaces Orion Flash 2 on a new engine. It is a 552-billion-parameter mixture-of-experts model with an asymmetric design — 8 billion parameters active per token during prefill and 16 billion during decode — that reads photos natively, runs a deliberate reasoning pass before answering, and keeps a roughly one-million-token context window, while answering faster and consuming less included usage than Orion 2.1.
Available in Plan mode on Pro and MAX.
More about Nox Orion Flash 2.1
Nox 1.3 Astra (Retired)
Deep reasoning with a one-million-token context — retired October 5, 2026
1.3 Astra was the deep-reasoning model of the Nox 1.3 generation. It was a 700-billion-parameter mixture-of-experts reasoning engine that activated 36.9 billion parameters per token, read a context window of roughly one million tokens, and ran a deliberate reasoning pass before every answer. It was retired on October 5, 2026, when its underlying engine was discontinued; Nox Orion 2.1 now covers deep multi-step work.
Retired on October 5, 2026. No longer available; Nox Orion 2.1 now covers deep multi-step work.
Nox Vela 2.5 (New)
The fast everyday model on every plan — now reads images
Vela 2.5 is the everyday model of the Nox family and the default for everyday chat on every plan, including Free. It replaced Vela 2.1 in November 2026 and keeps the Vela line's headline property — speed — while reading photos directly instead of handing them to a separate vision model.
Available on every plan, including Free. Long conversations (over 100,000 tokens) are MAX and above.
Nox Vela 2.1 (Retired)
Replaced in chat by Vela 2.5 — the fast model of the Nox 2.1 generation
Vela 2.1 was the everyday model of the Nox family and the one available on every plan, including Free. It built on every previous Vela generation with higher accuracy at the same speed, on a 30-billion-parameter hybrid mixture-of-experts engine that activates 3 billion parameters per token. On November 9, 2026 it was replaced in chat by Nox Vela 2.5; some voice and agent steps still run on Vela 2.1.
Replaced in chat by Nox Vela 2.5 on November 9, 2026. Some voice and agent steps still run on Vela 2.1.
Nox Orion 2 (Retired)
Retired — the first flagship of the Nox 2 generation
Orion 2 was the direct upgrade to 1.4 Orion and the first flagship of the Nox 2 generation. It kept everything 1.4 Orion could do — the same tools, thinking levels, and answer standard — while responding faster overall and costing less. It was an 824-billion-parameter mixture-of-experts model, 46 billion active per token, with a roughly one-million-token context window. On October 22, 2026 it was succeeded by Nox Orion 2.1, which runs the same engine with a new peer-review answer pipeline.
Retired on October 22, 2026 — succeeded by Nox Orion 2.1.
Nox Orion Flash 2 (Retired)
Retired — the first fast Orion, a 300B model
Orion Flash 2 was the first fast version of Orion. It was a 300-billion-parameter model built for speed that kept Orion's thinking depth intact, running the same deliberate reasoning pass while answering faster and consuming less included usage than Orion 2. On October 22, 2026 it was succeeded by Nox Orion Flash 2.1, which moves to a 552-billion-parameter mixture-of-experts engine with native photo input.
Retired on October 22, 2026 — succeeded by Nox Orion Flash 2.1.
Nox Vela 2 (Retired)
Retired — the Nox 2 everyday model, continued as Vela 2.1
Vela 2 was the everyday model of the Nox 2 generation and the one available on every plan, including Free. It ran on a 30-billion-parameter hybrid mixture-of-experts engine that activates 3 billion parameters per token, with a roughly 262,000-token context window. On October 22, 2026 the Vela 2 name was retired and the same model continued as Nox Vela 2.1, at the same price.
Retired on October 22, 2026 — continued as Nox Vela 2.1, since replaced by Nox Vela 2.5.
Nox 1.4 Orion (Retired)
Retired — the flagship of the Nox 1.4 generation
1.4 Orion was the first model of the Nox 1.4 generation: an 824-billion-parameter mixture-of-experts model, 46 billion active per token, with a roughly one-million-token context window. It reasoned through a deliberate internal pass before answering and sized each answer to the question, up to 9,000 characters. It is no longer selectable under this name; Nox Orion 2.1 is its successor.
Retired — succeeded by Nox Orion 2.1.
Nox 1.3 Nova (Retired)
Retired — the balanced model of the 1.3 family
1.3 Nova sat directly below Astra in the Nox 1.3 family. It was a hybrid-reasoning model: it decided per question whether to run a step-by-step reasoning pass or answer directly, which kept replies conversational in speed while staying structured in substance. It is no longer part of the active lineup; Nox Vela 2.5 and Nox Orion 2.1 cover its range.
Retired — no longer part of the active lineup.
Nox 1.3 Vela (Retired)
Retired — the fast model of the 1.3 family
1.3 Vela was the everyday model of the Nox 1.3 family and available on every plan, including Free. It ran on a 30-billion-parameter hybrid Mamba-Transformer mixture-of-experts engine that activated 3 billion parameters per token. Nox Vela 2 replaced it on September 23, 2026; the line continues today as Nox Vela 2.5.
Retired — the Vela line continues as Nox Vela 2.5.
Nox Light 1.1 (Retired)
Retired — Nox's previous-generation fast model
Light 1.1 was the previous-generation model behind everyday Nox conversations, tuned for low latency and concise answers. It was retired from Nox consumer products on September 23, 2026. Nox Vela 2.5 is its current successor, while Light remains available to developers through the Nox developer API.
Retired from consumer products. Nox Vela 2.5 is its current successor; Light remains available through the Nox developer API.
Nox Agent 1.2 (Retired)
Retired — Nox's former dedicated research mode
Nox Agent 1.2 was the dedicated research mode in Nox. It planned multi-step research runs and returned answers with cited resource cards. The research mode was discontinued on September 23, 2026; Plan mode with live web access is now the way to get cited research.
Retired on September 23, 2026. No longer available; use Plan mode with live web access for cited research.
Nox Voice 2 (New)
The most powerful Nox voice yet — expressive, interruptible, Plan mode in the call
Nox Voice 2 is the voice of Nox and the most powerful voice model Nox has ever shipped — more than 50% less cost per call than Nox Voice 1 Advance. It is built to sound like a person on a call rather than a narrator: the model writes short delivery directions into its own reply — warmth, a pause, a quiet tone, a light chuckle — and the speech engine performs them. It replies in whatever language you just used, mixing languages mid-sentence when you do, and it stops the moment you say something.
Available on every paid plan (Pro and MAX). Not included on Free.
Nox Voice 1 Advance (Retired)
Spoken Nox on paid plans — retired October 17, 2026
Nox Voice 1 Advance was Nox's second-generation spoken conversation system: speak, receive a focused spoken reply in under two seconds, and keep the exchange in the same chat thread. It was retired on October 17, 2026, and is no longer available; Nox Voice 2 replaced it.
Retired on October 17, 2026. No longer available; Nox Voice 2 is its successor.
More about Nox Voice 1 Advance
Nox Voice 1.2 (Retired)
Real-time voice beta — retired October 17, 2026
Nox Voice 1.2 (beta) was the first generation of spoken Nox: a live, low-latency speech loop with interruption support and a transcript saved to the chat. It was retired on October 17, 2026, and is no longer available; Nox Voice 2 replaced it.
Retired on October 17, 2026. No longer available; Nox Voice 2 is its successor.
Nox Thinking 1.1 (Retired)
Retired — Nox's first dedicated reasoning model
Nox Thinking 1.1 was the Nox 1.1 generation's dedicated reasoning model. It ran a deliberate reasoning pass before answering, trading speed for more structured, multi-step responses. It was retired in July 2026 and is no longer available.
Retired in July 2026. No longer available.