--- title: GPT-6.1 Terra is OpenAI's bet on cheap computer use description: OpenAI has launched GPT-6.1 Terra, a model built for cheaper, faster computer-use agents rather than top-tier intelligence. It understands on-screen content... canonical: https://daily.dev/posts/gpt-6-1-terra-is-openai-s-bet-on-cheap-computer-use-bc479nxfn twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: GPT-6.1 Terra is OpenAI's bet on cheap computer use | daily.dev og:description: OpenAI has launched GPT-6.1 Terra, a model built for cheaper, faster computer-use agents rather than top-tier intelligence. It understands on-screen content... og:url: https://daily.dev/posts/gpt-6-1-terra-is-openai-s-bet-on-cheap-computer-use-bc479nxfn og:image: https://api.daily.dev/og/posts/BC479NXFN.png og:image:alt: GPT-6.1 Terra is OpenAI's bet on cheap computer use og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # GPT-6.1 Terra is OpenAI's bet on cheap computer use **[Trends](https://daily.dev/sources/trends)** · 3 min read · 2 upvotes · 0 comments ## Summary OpenAI has launched GPT-6.1 Terra, a model built for cheaper, faster computer-use agents rather than top-tier intelligence. It understands on-screen content and takes actions across apps, browsers, and desktops. It's rolling out in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu plans, with API access coming next for high-volume agent workflows. OpenAI hasn't published pricing or benchmark numbers for the computer-use claims, so real-world performance is still untested. ## Content OpenAI's pitch for GPT-6.1 Terra is blunt: Astra-class results, a fraction of the cost. Per OpenAI, it matches GPT-6 Astra on Agents' Last Exam at 1/10 the price per token and finishes tasks about 3x faster. API prices are 50% below GPT-5.6 Terra. Amazon Bedrock's listing adds a 15-point jump over GPT-5.6 Terra on the same exam, a 1M-token context window, and text and image inputs. ## The hype is loud @gdb says computer use with Terra "has crossed a qualitative threshold where it feels faster at using a computer than what i can do unassisted." Sam Altman keeps it short: "6.1 Terra is here! It is good at a lot of things, but we are especially excited for this new era of computer use." @AlexFinn is the most excited of the bunch. He couldn't believe the numbers ("Ummm am I reading this correctly?"), then pointed at the marketing: visual reasoning, filling forms, editing videos, drafting proposals. His read is that this is OpenAI's first model built with personal agents in mind. He plans to plug it into Hermes. For coding he's staying on Astra with fast settings, so even the fan isn't switching everything over. A reshared post from @mattworkman says Terra (Ultra) is "INSANELY good" at 3D Blender computer use, and that it used only 3% of his Pro 200 plan. ## The sober reads GitHub is more careful. Terra is now generally available in Copilot, and early testing showed it solving more terminal and C# tasks than GPT-5.6 Terra. GitHub gives no figure for how big the improvement is. One writeup of the Copilot rollout tells developers to track pass rates, corrections needed, and review time instead of trusting reported gains. Good advice for any launch with a "1/10th the price" slide. It's also worth noticing who's claiming what. The headline comparisons come from OpenAI's own benchmark, and the "Astra-level" claim is "on some tests," in AlexFinn's words. Astra and Sol remain the options for harder coding work. ## Where you can get it Terra is in ChatGPT Work and Codex on Plus, Pro, Business, Enterprise, and Edu plans. It's in the API as gpt-6.1-terra, on Bedrock, and in Copilot across VS Code, Visual Studio, JetBrains, Xcode, and the CLI. The model slug appeared in the API and Microsoft's eval configs before launch day, so nobody was surprised. The bet is that cheap, fast screen-driving agents are now good enough to run all day. @ISanjayD's summary: "Computer-use agents just got cheaper." Whether they got reliable enough is what the next week of testing will show. ## Questions this post answers ### What is GPT-6.1 Terra from OpenAI? It's an OpenAI model positioned for fast, lower-cost computer use, meaning it understands what's on screen and takes actions across apps, browsers, and desktops. It's available in ChatGPT Work and Codex on Plus, Pro, Business, Enterprise, and Edu plans, with API access coming for high-volume agent workflows. OpenAI hasn't published pricing or benchmarks for its computer-use claims yet. _Anyone evaluating agent tooling can track real-world tests of Terra's computer-use claims on daily.dev._ ### Is GPT-6.1 Terra available through the OpenAI API? Not yet at launch, it's first available in ChatGPT Work and Codex, but OpenAI says it is also coming to the API and Codex for developers building apps and agents that interpret images, navigate software, and take action, targeting high-volume workflows. _Developers planning agent integrations can follow the API rollout for Terra on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 2 discussions and 249 comments across hackernews, x (as of 2026-10-27). **TL;DR:** The 50% price cut on computer-use tasks is the headline people care about, with many framing it as the thing that finally makes always-on agents affordable, while others push back on reliability, hidden infra costs, and security risks of cheaper autonomous clicking. GPT-6.1 Terra: Most of the conversation isn't about Terra's capabilities but about OpenAI's confusing model naming/tiering (Luna/Terra/Sol/Astra) and how to choose between models for a given task; a smaller side-thread debates how Terra's benchmarks compare to Opus 5.5. **Sentiment:** 27% positive · 45% mixed · 28% skeptical **The case for** - Halving the API price for computer use makes it feasible to leave agents running all day instead of just in demos. - Some developers find Terra a genuinely useful cost/speed middle option that fills a gap between Luna and Sol for agentic and API workloads. - A 1M-token context window at half price makes long-document and whole-codebase workflows much cheaper. - One developer called it a great release for cost-sensitive agentic production use via the API. - Faster task completion plus lower price is seen as what finally moves agents from demos to daily production use. **The pushback** - Cheap tokens don't help if agents quit tasks early or need retries, since cost-per-finished-task is what actually matters. - Many find the Luna/Terra/Sol/Astra naming and versioning scheme confusing, especially when newer smaller models outperform older larger ones. - Cheaper, more numerous agents browsing and clicking real pages raises prompt-injection and accountability risks that need tighter permissions, not looser ones. - Several argue there's no clear guidance on which model/effort level to use for which task, making model selection largely folklore. - Model token cost dropping doesn't reduce the underlying compute/VM cost of running browser sessions, so savings may be overstated. **By community** - hackernews (mixed): Dominated by a long, heated debate over OpenAI's confusing model naming/tiering rather than the release's actual merits, with a smaller side-debate on benchmark credibility versus Opus 5.5. - x (mixed): Replies mix genuine enthusiasm about cheaper, faster computer-use agents with skepticism about reliability, hidden infrastructure costs, and security implications of cheaper autonomous agents. **Hottest debate:** Whether the 50% price cut is a genuine breakthrough for always-on agents or a misleading headline that ignores retry costs, infrastructure bills, and security exposure. **Open questions** - Does Terra's task success rate hold up once the cost-per-call drops, or do retries eat the savings? - How should users decide which model/effort-level combination to use for a given task without official guidance? - How much of the 'computer use' capability is actually new versus already present in prior models' tool calls? - Does a newer smaller model (e.g., Terra 6.1) reliably beat an older larger model (e.g., Sol 6), and under what conditions? - Who is accountable when cheaper, more numerous agents act online on people's behalf? **Highlights** > @testingcatalog Terra at $1/$5 is the headline, but the number that matters in production is cost per finished task, retries and early stops included. Cheap tokens that quit quietly get expensive. > — [VishalSharai on x · 1 points](https://x.com/VishalSharai/status/2115030341098549316) > @ai_for_success Terra gives up on jobs too easily. > — [MarkMesenko on x](https://x.com/MarkMesenko/status/2114961290364088561) > @testingcatalog Cheaper computer use is mostly a capacity story for whoever hosts these agents. Each session is a real browser holding RAM and CPU for minutes, so the model bill shrinks while the VM bill doesn't. I'd budget per concurrent session, not per token, before scaling it up. > — [mohammadvazin on x](https://x.com/mohammadvazin/status/2115022642541625835) > @testingcatalog Copilot auto approving clearly safe requests sounds nice until the agent reads a planted AGENTS.md. Would keep npm install and anything touching agent config on manual, at least for now. > — [lotar\_vk on x](https://x.com/lotar_vk/status/2115025229261246601) > @testingcatalog Halving the price of computer use matters more than another benchmark win. Most agent tasks people actually want are boring clicks across five apps, and cost per task is what kept them stuck in demos. Once it's cheap enough to run all day, reliabilty becomes the only number > — [Kieren\_PB\_Day on x](https://x.com/Kieren_PB_Day/status/2115026290680160381) **Source threads** - [hackernews](https://news.ycombinator.com/item?id=50248124) · 311 points · 188 comments - [x](https://x.com/testingcatalog/status/2115017548651462990) · 0 points · 46 comments - [x](https://x.com/ai_for_success/status/2114941219835097377) · 15 points · 15 comments ## Similar posts on daily.dev - [GPT-6 Astra Is Here. Is AGI Actually Any Closer?](https://daily.dev/posts/gpt-6-astra-is-here-is-agi-actually-any-closer--nhqbauuhl) · Medium · 13 upvotes · 5 comments - [GPT-6 Astra: The AI That Doesn’t Just Answer. It Gets the Job Done](https://daily.dev/posts/gpt-6-astra-the-ai-that-doesn-t-just-answer-it-gets-the-job-done-zotuejbng) · Medium · 0 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#openai](https://daily.dev/tags/openai), [#openai-codex](https://daily.dev/tags/openai-codex) [View this post on daily.dev](https://daily.dev/posts/gpt-6-1-terra-is-openai-s-bet-on-cheap-computer-use-bc479nxfn) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"GPT-6.1 Terra is OpenAI's bet on cheap computer use","url":"https://daily.dev/posts/gpt-6-1-terra-is-openai-s-bet-on-cheap-computer-use-bc479nxfn","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/gpt-6-1-terra-is-openai-s-bet-on-cheap-computer-use-bc479nxfn"},"datePublished":"2026-10-26T18:08:04.265Z","dateModified":"2026-10-27T12:49:07.783Z","description":"OpenAI has launched GPT-6.1 Terra, a model built for cheaper, faster computer-use agents rather than top-tier intelligence. It understands on-screen content...","isAccessibleForFree":true,"articleSection":"Trends","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Trends","logo":"https://media.daily.dev/image/upload/s--ZfSp3asX--/f_auto,q_auto/v1780996004/logos/trends?_a=BAMAMiWQ0","url":"https://daily.dev/sources/trends"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/gpt-6-1-terra-is-openai-s-bet-on-cheap-computer-use-bc479nxfn","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,openai,openai-codex","timeRequired":"PT3M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Trends","item":"https://daily.dev/sources/trends"},{"@type":"ListItem","position":3,"name":"GPT-6.1 Terra is OpenAI's bet on cheap computer use"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/gpt-6-1-terra-is-openai-s-bet-on-cheap-computer-use-bc479nxfn#faq","mainEntity":[{"@type":"Question","name":"What is GPT-6.1 Terra from OpenAI?","acceptedAnswer":{"@type":"Answer","text":"It's an OpenAI model positioned for fast, lower-cost computer use, meaning it understands what's on screen and takes actions across apps, browsers, and desktops. It's available in ChatGPT Work and Codex on Plus, Pro, Business, Enterprise, and Edu plans, with API access coming for high-volume agent workflows. OpenAI hasn't published pricing or benchmarks for its computer-use claims yet. Anyone evaluating agent tooling can track real-world tests of Terra's computer-use claims on daily.dev."}},{"@type":"Question","name":"Is GPT-6.1 Terra available through the OpenAI API?","acceptedAnswer":{"@type":"Answer","text":"Not yet at launch, it's first available in ChatGPT Work and Codex, but OpenAI says it is also coming to the API and Codex for developers building apps and agents that interpret images, navigate software, and take action, targeting high-volume workflows. Developers planning agent integrations can follow the API rollout for Terra on daily.dev."}}]} ```