LLM Models

Regal offers multiple leading Large Language Models (LLMs) to power your AI agents for phone and SMS conversations. Each is optimized for different use cases – so the best choice depends on your goals, call type, complexity of conversation, and cost tolerance.

We recommend starting with OpenAI GPT-4o Mini, Regal's default model. Then, progressively upgrade to higher-performing or more specialized models (e.g., GPT-4.1 Mini, GPT-4.1, GPT-4o, or the new GPT-5 series) only if you find that your AI agent isn’t performing the call as well as you’d like and you’re willing to pay more.

GPT-4o Mini

Best for short, simple cost-sensitive use cases
(e.g., calls normally staffed out of low-cost locales)

Examples:

Confirmation
Scheduling
Collections calls
Qualification calls with < 15 questions (and not much branching)

✅ Pros: Fastest model, cost-effective, great for straightforward tasks/conversations. Included in Regal base AI voice agent minute pricing
❌ Cons: Slightly weaker at reasoning; use a custom action for tasks like math

GPT-4.1 Mini

Best for short, simple conversations where strict task-instruction following is needed (improvisation not valued)

Examples:

Confirmation
Scheduling
Collections calls
Qualification calls with < 25 questions

✅ Pros: More reliable and consistent at following task instructions and function calling; updated knowledge through 2024
❌ Cons: Slightly worse at improvisation; slightly higher cost

GPT-4.1 / GPT-4o

Best for longer, more complex calls or those requiring math / reasoning

We've found GPT-4.1 better at following prompt instructions and function calling, while GPT-4o generates higher-quality, more natural conversation.

Examples:

Long qualification calls with 10+ steps or conditional branches
Open-ended support calls
Calls with multiple or iterative custom actions
Calls requiring complex reasoning (e.g., dynamic quote estimates)

✅ Pros: Handles conditional logic, multi-step functions, and nuanced dialogue better
❌ Cons: More expensive; adds slight latency

GPT-5 Series

Best for advanced reasoning, memory retention, and multi-turn context

OpenAI’s latest GPT-5 models extend GPT-4o’s strengths with deeper reasoning, improved memory, and more human-like conversational flow.

GPT-5 Nano: Ultra-lightweight, optimized for high-volume, low-cost interactions (e.g., confirmations, reminders)
GPT-5 Mini: Balanced option for mixed workloads requiring stronger reasoning at moderate cost
GPT-5: Full model, ideal for high-stakes or complex conversational tasks requiring domain knowledge, multi-step reasoning, and long-context recall
GPT-5.1: General-purpose conversational model with stable reasoning and consistent performance across multi-turn conversations, suitable for most production AI Agents that require reliable instruction following and moderate context depth.
GPT-5.2: Advanced conversational model optimized for complex workflows, offering stronger reasoning robustness, improved instruction adherence, and better handling of long, multi-step conversations with higher context demands.
GPT-5.4: Advanced model for complex professional workflows, offering strong reasoning, instruction following, tool use, and long-context performance for sophisticated, high-value AI Agent interactions.
GPT-5.4 Mini: Faster, more cost-efficient version of GPT-5.4, designed for high-volume workloads that still require strong reasoning, reliable function calling, and responsive conversational performance.
GPT-5.2 Nano: This does not appear to be an official OpenAI model. If you meant GPT-5.4 Nano: Lightweight GPT-5.4 model optimized for speed and cost on simple, high-volume tasks such as classification, data extraction, routing, and structured information capture.
GPT-5.6 Terra ("mini"): Balanced GPT-5.6 model combining strong reasoning with moderate cost and latency, suitable for production AI Agents handling multi-step conversations, tool calls, conditional workflows, and more nuanced customer requests.
(**Regal top pick**) GPT-5.6 Luna ("nano"): Most cost-efficient GPT-5.6 model, optimized for fast, high-volume workloads with well-defined instructions, including qualification, classification, information capture, and straightforward customer interactions.

When to Consider Claude

If you’ve tried OpenAI GPT-4o Mini / 4.1 / 5 Mini and need stronger performance around memory, creativity, or nuanced language, consider Anthropic’s Claude series:

Claude 3.5 Haiku: Fast, lightweight conversations
Claude 3.7 Sonnet: Longer, more complex reasoning tasks
Claude 4.0 Sonnet: Excels at subtle emotional understanding and creative synthesis

✅ Pros: Exceptional interpretability and safety; strong long-context memory
❌ Cons: Slightly slower response time and higher cost

When to Consider Gemini

Google’s Gemini series now includes the 2.5 and 3 lines:

Gemini 2.5 Flash / Flash Lite: Optimized for ultra-low latency and low-cost deployments, ideal for voice agents needing fast turn-taking
Gemini 3 Flash: Lightweight, cost-efficient conversational model optimized for high-volume workloads with simpler reasoning requirements, best suited for transactional or narrowly scoped conversations where throughput and cost efficiency are priorities.

✅ Pros: Excellent multimodal performance and web-connected reasoning
❌ Cons: Slightly less predictable function calling

Open Weight Model (in Beta)

Regal can provide an open source model option on request, on a case by case basis. We use GLM-4.7, an open-weight large language model developed by Zhipu AI (zai-org), available at https://huggingface.co/zai-org/GLM-4.7."

Switching Between Models

When switching models, expect to make small prompt adjustments. For example, GPT-4o Mini may require more explicit phrasing, while GPT-5 or Claude 4.0 can infer intent more flexibly. Always test before full rollout.

Still not sure? Reach out – we’re happy to help tune your agent for best performance.

Want Regal to offer a different LLM? Just reach out – we’re constantly evaluating new models and onboarding those with differentiated performance.

Want to bring your own LLM? Contact us – we’ll help you integrate it into Regal.