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 is its successor.
Architecture: mixture-of-experts, 824 billion parameters total, 46 billion active per token — the 700-billion-parameter backbone of 1.3 Astra extended with roughly 100 billion parameters of additional trainable components.
Length: earlier Nox models answered under a fixed 4,000-character cap. Lifting that cap alone raised HealthBench Hard by 8.4 ± 2.0 points on a 150-example paired test, almost entirely on completeness. Orion scaled length to the question instead — short for simple questions, up to 9,000 characters for multi-part ones.
HealthBench Hard, Nox self-run, official open protocol, all 1,000 examples: 46.4% (±0.8) at Ultra on September 5, 2026; 53.3% (±1.0) at Hyper Ultra on September 11, 2026. 1.3 Astra on the identical protocol: 34.8%. These remain the most recent published Orion results; they have not been re-measured under the Orion 2 name.
Hyper Ultra was Orion’s sixth and deepest thinking level, in Plan mode on MAX. Same model, several passes over one question — draft, self-written rubric, self-grade, revision — about five to six minutes per answer. On a fixed 150-prompt development subset the lower levels scored Light 36.2%, Balanced 36.8%, Extra 36.9%, Max 38.5%.
Multimodal and tool-using: up to ten photos per message; live web access in Plan mode with up to ten search-and-read rounds per answer, every source cited; inline health and anatomy image generation; thinking time shown on every answer and a live step timeline in Plan mode.
1.4 Orion was succeeded by Nox Orion 2 on September 23, 2026 — the same capabilities with faster overall responses, higher accuracy, and a lower price.
Capabilities
- 824-billion-parameter mixture-of-experts model, 46 billion active per token
- ~1,000,000-token context window
- Deliberate internal reasoning pass before every answer, thinking time shown
- Six thinking levels, Light to Hyper Ultra
- Dynamic answer length, up to a 9,000-character cap
- Photo input — up to ten images per message
- Live web access in Plan mode — up to ten search-and-read rounds per answer, every source cited
Model details
- Architecture — Mixture-of-experts — 824 billion parameters total, 46 billion active per token
- Context window — ~1,000,000 tokens
- Answer length — Dynamic — sized to the question, up to a 9,000-character hard cap
- HealthBench Hard — 46.4% (±0.8) at Ultra, September 5, 2026; 53.3% (±1.0) at Hyper Ultra, September 11, 2026 — Nox's own self-runs, official open protocol, all 1,000 examples; not independently verified
- Raw evidence — Every answer, every grader verdict, aggregate files and SHA-256 checksums are published: Ultra at https://www.getnox.ai/data/healthbench/orion/README.md, Hyper Ultra at https://www.getnox.ai/data/healthbench/orion-hyper-ultra/README.md
- Status — Retired on September 23, 2026 — succeeded by Nox Orion 2
Availability: Retired — succeeded by Nox Orion 2.