Models
| Model | ID | Available in |
|---|---|---|
| Nimbus 2 Apex | nimbus-2-apex | API, online studio, offline studio |
| Nimbus 1.1 Prime-EE | — | Offline studio |
Nimbus 2 Apex
Nimbus 2 Apex is a LoRA adapter over Qwen3-8B, trained to follow instructions closely. It reasons before it answers, and two settings control how it does so.
- Reasoning level (1 to 20, default 5). Sets how long the model reasons before answering. The reasoning stops at a ceiling of two and a half times the target below.
- Temperature (1 to 10, default 5). Lower values give more predictable answers, higher values more varied ones.
| Level | Target length of reasoning |
|---|---|
| 1 | about 15 words |
| 5 | about 170 words |
| 10 | about 470 words |
| 15 | about 870 words |
| 20 | about 1,340 words |
Other properties:
- Languages: English and Persian.
- Input: text, and images read by the image module apex-imagexo-v1 (API reference).
- Context online: 40,960 tokens per request, including tools, reasoning and answer.
- Tools: function calling, and web search run by AhoosAI (API reference).
- Online, it runs on an NVIDIA T4 graphics card with PML-sRUN.
- Weights: huggingface.co/AhoosAI/nimbus-2-apex. Measurements: Nimbus 2 Apex — the measurements.
Nimbus 1.1 Prime-EE
A behaviour adapter over Qwen2.5-Coder-7B-Instruct. It focuses on what programmers often get wrong by habit: right-to-left layouts, accessibility and hardened containers, and it answers in the language of the question. It runs in the offline studio and is not served by the API.
- Weights: huggingface.co/AhoosAI/nimbus-1-1-prime-ee. Measurements: Nimbus 1.1 Prime-EE — the measurements.
Retired models
The Nimbus RZA family has been retired and is no longer available.