Which AI Model Should a Support Agent Run On in 2026? GPT-5.6, GPT-5.4, Claude and Groq

Pick the model by the job, not the bill: on Macha every ticket costs the same whichever model answers it (about $0.40 a ticket, $299 a month for 750 tickets), so the choice comes down to capability and speed. GPT-5.4 Mini suits fast, high-volume lookups, GPT-5 and GPT-5.4 suit nuanced or strict agents, and GPT-5.6 Terra, added on July 27, 2026, is Macha's flagship for the hardest multi-step work.

That billing rule is the whole reason this question changed. You're charged per ticket: one thread between Macha and one person, charged once however many messages, tool calls or follow-ups it takes. The model an agent runs on doesn't move that number. What's left is the question worth asking anyway: which model does this agent's job best?

Which models can a Macha support agent run on?

The full picker, and where each model fits:

GPT-5.4, GPT-5 and GPT-5.4 Mini are now the previous generation of OpenAI models in the picker. They stay there, and agents already running on them keep that model until you change it. Nothing gets migrated behind your back.

For context on the trajectory: GPT-5.4 launched in Macha on April 23 as OpenAI's most capable model, and at the time it was expensive enough to ration. Two rounds of price cuts later, Macha moved to per-ticket billing and took cost out of the question. Then the GPT-5.6 pair arrived on top.

Does the model change what a ticket costs?

No. Under per-ticket billing, the arithmetic that used to fill this section disappears:

That's the whole table. Pricing is $299 a month for 750 tickets and scales from there, about $0.40 a ticket on every tier, and tool calls and sub-agent delegation sit inside the same ticket. The capable models didn't get cheaper so much as the budget argument for holding an agent back went away.

If you run low, there are no overages: agents pause rather than billing on. You move up a tier or add a top-up. See the pricing page for current tiers.

How do you pick a model for a support agent?

The industry framing, borne out across model-selection guides from Microsoft's Foundry team and support-focused write-ups like Cobbai's LLM evaluation guide, is that you're trading off three things: capability, latency, and cost. On Macha cost drops out, so two are left. There's no single best model; there's a best model for a given agent's job. Here's how we'd map the OpenAI options.

Use GPT-5.4 Mini for the fast, routine majority

For much of support work, answering "where's my order," explaining a policy, looking up an account, triaging and routing, GPT-5.4 Mini is a sound call. It's fast, which is what your busiest agents actually need, and OpenAI lists a 400,000-token context window, enough to hold long ticket histories and knowledge-base context without truncating. Mini-class models are the usual pick for high-volume, latency-sensitive support, and that matches what we see. We wrote a deeper piece on exactly this: GPT-5.4 Mini for customer support.

Step up to GPT-5 or GPT-5.6 Luna when replies need more nuance

GPT-5 is the balanced middle: better at nuanced, judgment-heavy replies than Mini, and still quick enough to run at scale. GPT-5.6 Luna now sits in the same slot from the newer generation, with the same roughly 1M-token context as Terra. Good candidates are agents that draft customer-facing prose where tone and reasoning matter: apology-and-remedy responses, technical explanations, anything where a flat or literal answer would read poorly. There's no premium over Mini to justify; the only cost is a little latency.

Reserve GPT-5.4 or GPT-5.6 Terra for strict, complex, high-stakes agents

GPT-5.4 is the default for new agents, and its real edge over GPT-5 is instruction-following: agents with strict "do this exactly once" rules, complex branching logic, or many tools where the model has to pick the right one in the right order. GPT-5.6 Terra goes a step further on multi-step reasoning and tool selection, and it's the one to try first when an agent keeps skipping a step or calling a tool twice. These are the models for billing and refund flows, multi-tool runbooks, and orchestration agents that delegate to sub-agents, the places where a wrong action is expensive.

A common pattern is tiering: a Mini-powered front-line agent handles the routine majority and delegates the hard cases to a Terra or GPT-5.4 specialist sub-agent. That keeps the queue moving and puts the careful thinking only where it's needed, and the whole chain is still one ticket. Because Macha lets each agent and each sub-agent set its own model, you can build that tier directly. The agents documentation covers per-agent model selection and delegation.

When should you use Claude Sonnet or a Groq model instead?

If you've read any of the cross-provider round-ups, like the LLM API pricing comparisons that stack OpenAI against Claude, Gemini and Llama, you'll know choosing a model is rarely an OpenAI-only decision. It isn't in Macha either. The picker also includes Anthropic's Claude Sonnet 5 and 4.5 and Groq-hosted open models (GPT OSS 120B and GPT OSS 20B), and they earn their place for different jobs:

The pattern in support-LLM guides is that mini-class OpenAI models win on all-round value, Claude does well on instruction adherence and long-context reasoning, and the fastest providers win when latency dominates. So if an agent's job is sensitive handling, refund apologies, complaints, anything where tone carries the resolution, A/B test Claude Sonnet against GPT-5 on a few real tickets before you commit. And for a dead-simple, high-frequency lookup agent, a Groq model can beat GPT-5.4 Mini on speed. Match each agent to the provider that fits its job, with capability and latency driving the choice.

Where do you set an agent's model?

The model dropdown in the top-right of a conversation only changes the model for that one chat. It's there for quick experiments: try a reply on GPT-5.6 Terra, compare it to Mini, decide. If you want an agent to permanently run on a different model, that lives in the agent's configuration, and the org-wide default lives in Settings. The dropdown shows a toast saying as much, so nobody assumes a one-off switch stuck. It's a small thing that prevents a real "why is my agent still on the old model" confusion.

When shouldn't you reach for the most capable model?

A flat per-ticket price is not a reason to put every agent on Terra. A few honest caveats:

• More capable isn't always better for support. For fast, friendly, high-volume chat, a lighter model is often the better customer experience because it's quicker. Capability you don't need just adds latency. Don't upgrade an agent that's already answering correctly.
• Match the model to the cost of being wrong. A wrong refund or a mis-routed escalation is expensive in a way no model choice is. That's the axis to optimise now that price isn't one.
• Vision and tooling vary by model. Macha's Read Attachment tool reads images, PDFs and documents on Zendesk tickets, and OpenAI lists image input for both GPT-5.4 and GPT-5.4 Mini. If an agent needs to read receipts or screenshots, confirm the model you're switching to accepts images before you move it, especially for the Groq-hosted text models.
• You're not buying tokens. Macha's unit is the ticket, across every provider; token economics differ by model but never reach your invoice. Don't try to reverse-engineer an OpenAI bill from it.
• Test before you roll out. Use the per-conversation dropdown to compare a real ticket on two models before you change an agent's default.

The healthiest setup for most teams is still a light model on the routine lookups, GPT-5 or Luna on the customer-facing agents that write a lot of prose, and GPT-5.4 or Terra reserved for the strict, complex few. Nothing about your bill argues against the upper tiers.

FAQ

Does the model I pick change what I pay? No. Macha bills per ticket, one thread charged once, whichever model the agent runs on. Pricing starts at $299 a month for 750 tickets, about $0.40 a ticket.

Do I have to do anything to switch an agent up a tier? No. Change its model in the agent's configuration and nothing else moves. There's no migration and no billing consequence.

Is a billed ticket the same as a resolved ticket? No. A ticket is one conversation, one thread with one person, charged once whether it ends resolved or handed to a human. Macha is automation and orchestration, so outcomes vary; the meter counts conversations, not wins.

Which model should most of my agents use? GPT-5.4 Mini for fast routine lookups, GPT-5 or GPT-5.6 Luna for nuanced customer-facing replies, and GPT-5.4 or GPT-5.6 Terra for strict, complex or high-stakes agents. See the GPT-5.4 Mini deep dive for the high-volume case.

Is OpenAI the only provider, or can I use Claude? Macha isn't OpenAI-only. The model picker also includes Anthropic's Claude Sonnet 5 and 4.5, strong on tone and long-document reasoning, and Groq-hosted open models (GPT OSS 120B, GPT OSS 20B) for very fast responses. For tone-sensitive replies, test Claude Sonnet against GPT-5 on real tickets; for dead-simple high-frequency lookups, a Groq model can be the fastest option.

How do I change an agent's model permanently? In the agent's configuration card, not the conversation dropdown (which only affects the current chat). The org-wide default lives in Settings. The docs walk through it.

What is GPT-5.6 Terra? GPT-5.6 Terra is the OpenAI model Macha added on July 27, 2026 as its flagship, with a roughly 1M-token context window and sharper multi-step reasoning and tool selection. GPT-5.6 Luna arrived the same day as the faster mid-tier option.

Try it

Macha runs your AI agents on top of the help desk you already use, Zendesk, Freshdesk, Gorgias, Front, HubSpot or Intercom, and lets you pick the right model for each agent's job. If you've been holding a smart agent back on cost, there's no longer a reason to. Start a free trial ($50 of usage, no credit card required), or read more about running Macha on Zendesk.

Written by Abbas (Customer Support & AI, Macha) · Reviewed by Ankeet Guha (Co-founder & CTO) · Published 2026-06-24 · Last updated 2026-09-24.