Introducing GPT-5.5

We’re releasing GPT‑5.5, our smartest and most intuitive to use model yet, and the next step toward a new way of getting work done on a computer.

GPT‑5.5 understands what you’re trying to do faster and can carry more of the work itself. It excels at writing and debugging code, researching online, analyzing data, creating documents and spreadsheets, operating software, and moving across tools until a task is finished. Instead of carefully managing every step, you can give GPT‑5.5 a messy, multi-part task and trust it to plan, use tools, check its work, navigate through ambiguity, and keep going.

The gains are especially strong in agentic coding, computer use, knowledge work, and early scientific research—areas where progress depends on reasoning across context and taking action over time. GPT‑5.5 delivers this step up in intelligence without compromising on speed: larger, more capable models are often slower to serve, but GPT‑5.5 matches GPT‑5.4 per-token latency in real-world serving, while performing at a much higher level of intelligence. It also uses significantly fewer tokens to complete the same Codex tasks, making it more efficient as well as more capable.

We are releasing GPT‑5.5 with our strongest set of safeguards to date, designed to reduce misuse while preserving access for beneficial work. We evaluated this model across our full suite of safety and preparedness frameworks, worked with internal and external redteamers, added targeted testing for advanced cybersecurity and biology capabilities, and collected feedback on real use cases from nearly 200 trusted early-access partners before release.

Today, GPT‑5.5 is rolling out to Plus, Pro, Business, and Enterprise users in ChatGPT and Codex, and GPT‑5.5 Pro is rolling out to Pro, Business, and Enterprise users in ChatGPT. API deployments require different safeguards and we are working closely with partners and customers on the safety and security requirements for serving it at scale. We'll bring GPT‑5.5 and GPT‑5.5 Pro to the API very soon.

Model	Terminal-Bench 2.0	Expert-SWE (Internal)	GDPval (wins or ties)	OSWorld-Verified	Toolathlon	BrowseComp	FrontierMath Tier 1–3
GPT-5.5	82.7%	73.1%	84.9%	78.7%	55.6%	84.4%	51.7%
GPT-5.4	75.1%	68.5%	83.0%	75.0%	54.6%	82.7%	47.6%
GPT-5.5 Pro	-	-	82.3%	-	-	90.1%	52.4%
GPT-5.4 Pro	-	-	82.0%	-	-	89.3%	50.0%
Claude Opus 4.7	69.4%	-	80.3%	78.0%	-	79.3%	43.8%
Gemini 3.1 Pro	68.5%	-	67.3%	-	48.8%	85.9%	36.9%

OpenAI is building the global infrastructure for agentic AI, making it possible for people and businesses around the world to get work done with AI. Over the past year, we’ve seen AI dramatically accelerate software engineering. With GPT‑5.5 in Codex and ChatGPT, that same transformation is beginning to extend into scientific research and the broader work people do on computers.

Across these domains, GPT‑5.5 is not just more intelligent; it is more efficient in how it works through problems, often reaching higher-quality outputs with fewer tokens and fewer retries. On Artificial Analysis's Coding Index, GPT‑5.5 delivers state-of-the-art intelligence at half the cost of competitive frontier coding models.

Agentic Coding and Computer Use

GPT‑5.5 is our strongest agentic coding model to date. On Terminal-Bench 2.0, which tests complex command-line workflows requiring planning, iteration, and tool coordination, it achieves a state-of-the-art accuracy of 82.7%. On SWE-Bench Pro, which evaluates real-world GitHub issue resolution, it reaches 58.6%, solving more tasks end-to-end in a single pass than previous models. On Expert-SWE, our internal frontier eval for long-horizon coding tasks with a median estimated human completion time of 20 hours, GPT‑5.5 also outperforms GPT‑5.4.

Across all three evals, GPT‑5.5 improves on GPT‑5.4’s scores while using fewer tokens.

The model’s coding strengths show up especially clearly in Codex where it can take on engineering work ranging from implementation and refactors to debugging, testing, and validation. Early testing suggests GPT‑5.5 is better at the behaviors real engineering work depends on, like holding context across large systems, reasoning through ambiguous failures, checking assumptions with tools, and carrying changes through the surrounding codebase.

Beyond benchmarks, early testers said GPT‑5.5 shows a stronger ability to understand the shape of a system: why something is failing, where the fix needs to land, and what else in the codebase would be affected.

“The first coding model I’ve used that has serious conceptual clarity.”
— Dan Shipper, Founder and CEO of Every

After launching an app, he spent days debugging a post-launch issue before bringing in one of his best engineers to rewrite part of the system. To test GPT‑5.5, he effectively rewound the clock: could the model look at the broken state and produce the same kind of rewrite the engineer eventually decided on? GPT‑5.4 could not. GPT‑5.5 could.

“It genuinely feels like I’m working with a higher intelligence, and there’s almost a sense of respect.”
— Pietro Schirano, CEO of MagicPath

Pietro Schirano saw a similar step change when GPT‑5.5 merged a branch with hundreds of frontend and refactor changes into a main branch that had also changed substantially, resolving the work in one shot in about 20 minutes.

Senior engineers who tested the model said GPT‑5.5 was noticeably stronger than GPT‑5.4 and Claude Opus 4.7 at reasoning and autonomy, catching issues in advance and predicting testing and review needs without explicit prompting. In one case, an engineer asked it to re-architect a comment system in a collaborative markdown editor and returned to a 12-diff stack that was nearly complete. Others said they needed surprisingly little implementation correction and felt more confident in GPT‑5.5’s plans compared with GPT‑5.4.

One engineer at NVIDIA who had early access to the model went as far as to say: "Losing access to GPT‑5.5 feels like I've had a limb amputated."

"GPT-5.5 is noticeably smarter and more persistent than GPT-5.4, with stronger coding performance and more reliable tool use. It stays on task for significantly longer without stopping early, which matters most for the complex, long-running work our users delegate to Cursor."

Beyond Benchmarks: Everyday Work and Science

The same strengths that make GPT‑5.5 great at coding also make it powerful for everyday work on a computer. Because the model is better at understanding intent, it can move more naturally through the full loop of knowledge work: finding information, understanding what matters, using tools, checking the output, and turning raw material into something useful.

In Codex, GPT‑5.5 is better than GPT‑5.4 at generating documents, spreadsheets, and slide presentations. Alpha testers said it outperformed past models on work like operational research, spreadsheet modeling, and turning messy business inputs into plans. When combined with Codex’s computer use skills, GPT‑5.5 brings us closer to the feeling that the model can actually use the computer with you: seeing what’s on screen, clicking, typing, navigating interfaces, and moving across tools with precision.

Teams at OpenAI are already using these strengths in real workflows. Today, more than 85% of the company uses Codex every week across functions including software engineering, finance, communications, marketing, data science, and product management. In Comms, the team used GPT‑5.5 in Codex to analyze six months of speaking request data, build a scoring and risk framework, and validate an automated Slack agent so low-risk requests could be handled automatically while higher-risk requests still route to human review. In Finance, the team used Codex to review 24,771 K-1 tax forms totaling 71,637 pages, using a workflow that excluded personal information and helped the team accelerate the task by two weeks compared to the prior year. On the Go-to-Market team, an employee automated generating weekly business reports, saving 5-10 hours a week.

In ChatGPT, GPT‑5.5 Thinking unlocks faster help for harder problems, with smarter and more concise answers to help you move through complex work more efficiently. It excels at professional work like coding, research, information synthesis and analysis, and document-heavy tasks, especially when using plugins.

In GPT‑5.5 Pro, early testers are seeing a significant step up in both the difficulty and quality of work ChatGPT can take on, with latency improvements that make it much more practical for demanding tasks. Compared to GPT‑5.4 Pro, testers found GPT‑5.5 Pro’s responses significantly more comprehensive, well-structured, accurate, relevant, and useful, with especially strong performance in business, legal, education, and data science.

GPT‑5.5 reaches state-of-the-art performance across multiple benchmarks that reflect this kind of work. On GDPval, which tests agents’ abilities to produce well-specified knowledge work across 44 occupations, GPT‑5.5 scores 84.9%. On OSWorld-Verified, which measures whether a model can operate real computer environments on its own, it reaches 78.7%. And on Tau2-bench Telecom, which tests complex customer-service workflows, it reaches 98.0% without prompt tuning. GPT‑5.5 also performs strongly across other knowledge work benchmarks: 60.0% on FinanceAgent, 88.5% on internal investment-banking modeling tasks, and 54.1% on OfficeQA Pro.

“GPT-5.5 delivers the sustained performance required for execution-heavy work. Built and served on NVIDIA GB200 NVL72 systems, the model enables our teams to ship end-to-end features from natural language prompts, cut debug time from days to hours, and turn weeks of experimentation into overnight progress in complex codebases. It’s more than faster coding—it’s a new way of working that helps people operate at a fundamentally different speed.”

GPT‑5.5 also shows gains on scientific and technical research workflows, which require more than answering a hard question. Researchers need to explore an idea, gather evidence, test assumptions, interpret results, and decide what to try next. GPT‑5.5 is better at persisting across that loop than other models.

Notably, GPT‑5.5 shows a clear improvement over GPT‑5.4 on GeneBench(opens in a new window), a new eval focusing on multi-stage scientific data analysis in genetics and quantitative biology. These problems require models to reason about potentially ambiguous or errorful data with minimal supervisory guidance, address realistic obstacles such as hidden confounders or QC failures, and correctly implement and interpret modern statistical methods. The model’s performance is striking in light of the fact that tasks here often correspond to multi-day projects for scientific experts.

Similarly, on BixBench(opens in a new window), a benchmark designed around real-world bioinformatics and data analysis, GPT‑5.5 achieved leading performance among models with published scores. The model’s scientific capabilities are now strong enough to meaningfully accelerate progress at the frontiers of biomedical research as a bona fide co-scientist.

In another example, an internal version of GPT‑5.5 with a custom harness helped discover a new proof(opens in a new window) about Ramsey numbers, one of the central objects in combinatorics. Combinatorics studies how discrete objects fit together: graphs, networks, sets, and patterns. Ramsey numbers ask, roughly, how large a network has to be before some kind of order is guaranteed to appear. Results in this area are rare and often technically difficult. Here, GPT‑5.5 found a proof of a longstanding asymptotic fact about off-diagonal Ramsey numbers, later verified in Lean. The result is a concrete example of GPT‑5.5 contributing not just code or explanation, but a surprising a