LLM
Phantom Brain
Phantom Brain is Neuratel's own large language model, running on dedicated H100 infrastructure. It's included with your Neuratel plan — no external API key needed.
No API key required — uses Neuratel's own hardware.
Specifications
| Property | Value |
|---|---|
| Model name | "phantom" |
| Average latency | 187ms |
| Max completion tokens | 32,768 |
| Arabic support | Native |
Configuration
Code
Fields
| Field | Type | Default | Description |
|---|---|---|---|
model | string | "phantom" | Model identifier |
instructions | string | "You are a helpful AI assistant." | System prompt |
system_timezone | string | "UTC" | Timezone context |
temperature | float (0–2) | 0.8 | Response randomness |
max_completion_tokens | int (1–32768) | 4096 | Max response length |
top_p | float (0–1) | 1.0 | Nucleus sampling |
parallel_tool_calls | bool | true | Allow concurrent tool calls |
tool_choice | auto/required/none | auto | Tool use control |
max_tool_steps | int (1–20) | 3 | Max tool call iterations per turn |
Full Phantom Stack
For maximum performance on owned infrastructure — especially for Arabic — combine all three Phantom providers:
Code
This configuration makes zero external API calls — all inference runs on Neuratel's own hardware.
Related
- Phantom Speech — proprietary TTS
- Deepgram — recommended STT provider
Last modified on