Amp puts AI agents to work beyond code with OpenAI models
Amp is expanding its AI agent beyond code generation into writing, design, and real-time agent management, and OpenAI models account for the majority of tokens used in the product. The announcement matters because it shows a frontier coding agent treating model improvements as a trigger to widen the agent's job — from deciding what to build through shipping and post-launch marketing — rather than only accelerating code output.
The details come from an OpenAI customer story published on 2026-10-14, built around an interview with Quinn Slack, Amp's CEO and a member of its technical staff. The piece is a vendor case study rather than a technical paper: it names models and product features but publishes no benchmarks, latency figures, or pricing.
Amp is a frontier AI agent that is moving past code
Amp is described as building a frontier AI agent that helps developers turn ambitious ideas into working software. According to the post, each improvement in the underlying models is an opportunity for Amp to expand what the agent can do — from helping developers decide what to build, to writing code, to designing pages.
Amp chooses and recommends models based on what works best for its users. OpenAI models account for the majority of tokens used in Amp, according to the post.
Quinn Slack attributes part of that choice to transparency around safeguards:
"OpenAI is very transparent, and when there are safeguards applied, users know and they understand. That makes it really easy to use OpenAI models. You can do all of your work in Amp with OpenAI models and be really happy."
He also points to consistency across model generations as a reason users stay:
"Our users really appreciate the consistency with which OpenAI models have been getting better and better."
Sign in with ChatGPT lowers the perceived cost of long agent runs
Token cost becomes a constraint as builders hand agents more work. Quinn Slack describes some Amp users consuming 10 billion tokens a week, and argues that watching every increment of usage can hold people back from discovering what an agent is capable of.
Amp launched a Sign in with ChatGPT integration to address that. According to the post, the integration gives Amp users access to more tokens and OpenAI models, including GPT-6 Astra, and changes how willing users are to let an agent take on a larger task.
"Sign in with ChatGPT means that people can have that feeling where they're not counting every single token. And once you actually let the agent go and have a longer leash, then you realize just how much it can do."
Quinn Slack says he sees people using the agent to plan, decide what to build, and continue working after they ship.
GPT-6 Astra handles writing and design work for Amp's own team
Amp wants agents to help with all the work that goes into building and growing a product, and its own 20-person team uses them that way — including in marketing. The post quotes Quinn Slack saying the team has no person devoted to marketing, but does have an agent devoted to it. The agent is used to improve the experience for new users and to help more people discover Amp through search and other channels.
High-quality writing and design matter for that work, and GPT-6 Astra is used to produce documentation and design pages. Quinn Slack describes Amp's style standards as strict:
"We have a very unique design style and writing style. We're very picky about this, but we've been really happy with how GPT-6 Astra does."
The post also describes a personal experiment: Quinn Slack gave GPT-6 Astra the plans for his house and used it to create a 3D model in Blender, which he and his children then turned into games — adding a lazy river through the house or chasing garlic around it, inspired by their grandmother's dislike of garlic. He notes that this was not possible just a couple of months earlier.
Puck and GPT-Live-1 bring conversational control to agents
Amp recently announced Puck, a conversational AI assistant built into Amp that helps users launch and manage other agents. Users can ask Puck to start an agent to investigate a bug, coordinate several agents, or find a previous conversation about a technical problem.
A feature called Calling Puck, powered by GPT-Live-1, lets users talk to Puck in real time, including from a watch. Quinn Slack imagines people using it to keep their agents working while away from their computers, even while out for a run.
Ultrafast mode changes how people run agents
Amp has been testing Ultrafast mode with GPT-6 Astra. According to the post, the greater speed changes how people use agents. Today, users often switch between several agent tasks while waiting for each to finish; faster responses could let people juggle fewer agents, or give agents more time to check their own work before returning a result.
"People have not seen a model that's this good, that's this fast."
What comes next for Amp
For Amp, the stated next step is to bring more of the work before and after coding into the agent, so builders can carry an idea further without handling every step themselves.
"There's this unknown of what humans will create, and that's what we all get to see. It's cool to be playing a part in that."
What the announcement does and does not establish
The post establishes three named OpenAI models and features in use at Amp — GPT-6 Astra, GPT-Live-1, and Ultrafast mode — plus the Sign in with ChatGPT integration and the Puck assistant. It establishes that OpenAI models account for the majority of tokens used in Amp and that some Amp users consume 10 billion tokens a week.
It does not publish benchmark scores, latency measurements, pricing, or comparative evaluations for any of the named models. Those claims are absent from the source, and any figures beyond the ones quoted here are not supported by it.