HALO is built on a multi-model architecture. Rather than relying on a single AI model for everything, we select the right model for each task based on its strengths. This allows us to optimize for performance, reliability, and speed across different parts of the platform.
This article covers the models behind HALO's core features: agents, tools, guardrails, and search. Behind the scenes, HALO uses additional models for internal functions like automatic translation and platform-level safety, but those aren't covered here.
The Models
HALO currently uses two foundation models:
-
GPT 5.1 from OpenAI
-
Claude Haiku 4.5 from Anthropic
Each model has a distinct character. GPT 5.1 is adaptive and conversational—it handles open-ended tasks and ambiguity well. Haiku 4.5 is fast and precise—it excels at structured, well-defined tasks where consistency matters.
Where We Use GPT 5.1
GPT 5.1 powers the parts of HALO that require reasoning, flexibility, and natural conversation:
-
Agents — Powering conversations and task execution
GPT 5.1 is purpose-built for agentic work. It follows complex instructions without needing rigid prompt structures, adapts to ambiguous input gracefully, and produces responses that feel natural. When a task requires thinking through a problem or communicating with a user, GPT 5.1 handles it well.
Where We Use Haiku 4.5
Haiku 4.5 powers the parts of HALO that require speed, consistency, and precision:
-
Custom Guardrails — Evaluating inputs and outputs against your defined rules
-
Tool Step (LLM Interaction) — Single-pass LLM calls with a defined prompt and structured output
Haiku 4.5 is exceptionally fast and delivers consistent, deterministic outputs when given precise instructions. It's ideal for classification and decision-making tasks where you need the same input to produce the same result reliably. For high-volume operations where latency matters, Haiku 4.5 keeps things quick.
What This Means for You
You get the best of both worlds: agents that reason and communicate naturally, backed by fast, reliable infrastructure for everything that needs to run precisely and at scale.
As the AI landscape evolves, so will HALO. Our architecture is designed to adopt new models as they emerge—ensuring you're always building on a strong foundation.