Lara on Max and Agent Mode Just Moved to GPT-6 Luna

The AI landscape moves fast. One month a model feels cutting-edge, the next a new generation lands and quietly raises the bar on what “good enough” actually means for everyday work. We’ve been watching the GPT-6 family closely since OpenAI started rolling it out, and today we’re bringing the next step into ReadyTools.

Lara for Max subscribers is now powered by GPT-6 Luna.
Agent mode (available on Plus and Max) is also switching to GPT-6 Luna.

No settings to flip. No extra toggles. No plan changes required. If you’re already on Max, Lara is running the new model. If you use Agent mode on Plus or Max, that upgrade is live for you too.

Why this upgrade matters

GPT-6 Luna sits in the efficient, high-volume tier of OpenAI’s latest generation. It’s the successor to GPT-5.6 Luna — the same family lineage, but trained with the advances that came with GPT-6 Astra and then refined into the Sol and Luna variants. The practical differences show up in the places people actually notice when they work with AI every day.

Stronger instruction following and consistency.
You can give Lara longer, more detailed requests and expect her to stay on track across the whole conversation. Context holds better. Multi-step instructions are less likely to drift. When you’re iterating on a piece of writing, a plan, or a piece of analysis, the model is more reliable about remembering what you already decided three messages ago.

Better writing quality.
The outputs feel cleaner. Fewer odd phrasing choices, less unnecessary filler, clearer structure. Whether you’re drafting emails, blog posts, product copy, summaries, or internal docs, the results tend to land closer to something you can actually use with lighter editing.

Improved factual reliability.
OpenAI has reported meaningful reductions in the kinds of factual mistakes that used to appear in earlier generations. In real conversations that previously triggered corrections, the new models make fewer of those errors. That doesn’t mean zero mistakes — no model is perfect — but the baseline reliability is higher, which matters when you’re researching, summarizing sources, or using Lara for work that needs to be accurate.

Better at agent-style work.
Agent mode is where this upgrade is especially useful. When Lara is working with your Workspace pages, reading context, following multi-step flows, or chaining tools, the model’s improved ability to stay focused, respect boundaries, and handle longer sequences shows up. The earlier GPT-5.6 Luna version already improved instruction following and language consistency for Agent mode. GPT-6 Luna takes that further.

Large context, still practical.
Like its predecessor, GPT-6 Luna supports a very large context window (around a million tokens). That means you can bring bigger documents, longer conversation histories, or more complex briefs into a single request without the model losing the thread as quickly. For Max users who already lean on higher input limits, this remains a strong practical advantage.

Multimodal input.
It accepts text and images and returns text. Useful when you’re working with screenshots, diagrams, or reference images alongside your usual chat.

How this fits into ReadyTools

Lara is not a single-model chatbot. She’s the AI layer that sits across the platform — chat, Workspace, tools, search, and Agent workflows. We deliberately keep her on models that balance capability with the kind of usage patterns real people have: writing, research, planning, coding help, document work, and the multi-step tasks that Agent mode is built for.

Max is the plan designed for people who want the highest limits and access to the strongest models we make available. Plus is the plan that already unlocks Agent mode along with higher usage and the broader set of tools. Moving both Lara on Max and Agent mode (Plus + Max) onto GPT-6 Luna keeps those experiences current without fragmenting the product.

This is the same philosophy we’ve followed with previous model updates: when a new generation delivers clearer gains in the areas that matter for daily work, we bring it in. We did it when we moved Max onto earlier strong models. We did it when Agent mode received its GPT-5.6 Luna upgrade. We’re doing it again now.

What you don’t have to do

Nothing.
No new setting.
No model picker change.
No subscription upgrade required for this specific switch.

If you’re on Max, open Lara and keep working. If you use Agent mode on Plus or Max, the improved model is already in place. Existing conversations and workflows continue as usual — the intelligence underneath is simply stronger.

Looking ahead

The AI world isn’t slowing down. New model families arrive, older ones get refined, and the definition of “good everyday AI” keeps shifting upward. Our job is to keep ReadyTools moving with it so that the tools you already use — Lara, Agent mode, Workspace, and the rest of the platform — stay useful without forcing you to constantly relearn or reconfigure everything.

GPT-6 Luna is the latest step in that process. It brings better consistency, cleaner writing, stronger reliability, and improved agent behavior while keeping the practical strengths (speed, large context, everyday usability) that made the previous Luna generation work well inside ReadyTools.

We’re glad to have it live.
Go try it. Ask Lara something complex, hand Agent mode a real page or multi-step task, and see how it feels.

As always, thanks for building with us.

— The ReadyTools Team