LLM
OpenAI
OpenAI provides the gpt-4.1 and gpt-5.4 model families for agent LLM inference. Average latency: ~954ms.
Requires: OpenAI API key in Settings → Credentials
Models
| Model | Best For | Max Tokens |
|---|---|---|
gpt-4.1-mini | Recommended — fast and cost-effective | 128,000 |
gpt-4.1 | Complex reasoning, nuanced responses | 128,000 |
gpt-4.1-nano | Lowest latency and cost | 128,000 |
gpt-5.4 | Highest capability | 128,000 |
gpt-5.4-mini | Balanced performance | 128,000 |
gpt-5.4-nano | Ultra-fast, lightest model | 128,000 |
Configuration
Code
Fields
| Field | Type | Default | Description |
|---|---|---|---|
model | string | gpt-4.1 | Model to use |
instructions | string | "You are a helpful AI assistant." | System prompt |
system_timezone | string | "UTC" | Timezone for time-aware responses |
temperature | float (0–2) | 0.8 | Response randomness |
max_completion_tokens | int (1–128000) | 4096 | Max response length |
top_p | float (0–1) | 1.0 | Nucleus sampling |
frequency_penalty | float (-2 to 2) | null | Reduce repetition |
presence_penalty | float (-2 to 2) | null | Encourage new topics |
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 |
Example: Create Agent
Code
Related
- Building Your First Agent
- Groq — lower latency alternative
- Phantom Brain — no external key required
Last modified on