GPT-5.5: Everything We Know About OpenAI's Latest AI Model
On April 23, 2026, OpenAI released GPT-5.5 — codenamed internally as "Spud" — calling it "a new class of intelligence for real work." This is OpenAI's most capable production model to date, building on GPT-5.4 with major gains in agentic coding, computer use, knowledge work, and scientific research. Here's the complete guide.
What Is GPT-5.5?
GPT-5.5 (also known by its codename "Spud") is the direct successor to GPT-5.4, continuing the GPT-5 family that has progressed through GPT-5, GPT-5.1, GPT-5.2, and GPT-5.4. OpenAI describes it as their "smartest and most intuitive to use model yet" — one capable of taking complex, multi-part tasks and following them through to completion with minimal supervision.
The model excels at:
- Agentic coding — writing, debugging, and refactoring code across large systems
- Computer use — navigating software, clicking interfaces, operating tools autonomously
- Knowledge work — creating documents, spreadsheets, data analysis, and business research
- Scientific research — multi-stage biological and mathematical analysis
GPT-5.5 Benchmark Results
OpenAI published benchmark comparisons against GPT-5.4, Claude Opus 4.7, and Gemini 3.1 Pro:
| Benchmark | GPT-5.5 | GPT-5.4 | Claude Opus 4.7 | Gemini 3.1 Pro |
|---|---|---|---|---|
| Terminal-Bench 2.0 | 82.7% | 75.1% | 69.4% | 68.5% |
| SWE-Bench Pro | 58.6% | 57.7% | 64.3% | 54.2% |
| Expert-SWE (Internal) | 73.1% | 68.5% | — | — |
| GDPval (wins/ties) | 84.9% | 83.0% | 80.3% | 67.3% |
| OSWorld-Verified | 78.7% | 75.0% | 78.0% | — |
| BixBench (science) | 80.5% | 74.0% | — | — |
| CyberGym | 81.8% | 79.0% | 73.1% | — |
| ARC-AGI-2 | 85.0% | 73.3% | 75.8% | 77.1% |
| Tau2-bench Telecom | 98.0% | 92.8% | — | — |
GPT-5.5 achieves these results while matching GPT-5.4's per-token latency — a major engineering achievement for a significantly more capable model.
GPT-5.5 Variants
GPT-5.5 (Base)
The standard model available in ChatGPT and Codex for Plus, Pro, Business, and Enterprise users.
GPT-5.5 Pro
A higher-accuracy variant rolled out to Pro, Business, and Enterprise users. Early testers reported significantly more comprehensive, structured, and accurate responses — especially for business, legal, education, and data science tasks.
Key GPT-5.5 Pro benchmark highlights:
- BrowseComp: 90.1% (vs 84.4% base)
- FrontierMath Tier 4: 39.6% (vs 35.4% base)
- GeneBench: 33.2% (vs 25.0% base)
GPT-5.5 Thinking
Available in ChatGPT, this mode delivers "faster help for harder problems, with smarter and more concise answers." Ideal for professional work like coding, research, information synthesis, and document-heavy tasks.
GPT-5.5 Fast Mode (Codex)
Generates tokens 1.5x faster for 2.5x the cost — designed for latency-sensitive agentic workflows.
GPT-5.5 Pricing
API pricing (available from April 24, 2026):
- GPT-5.5: $5 per 1M input tokens / $30 per 1M output tokens
- GPT-5.5 Pro: $30 per 1M input tokens / $180 per 1M output tokens
- Batch/Flex: 50% of standard rate
- Priority processing: 2.5× standard rate
ChatGPT subscription access:
- Plus, Pro, Business, Enterprise: full GPT-5.5 access
- Free tier: not available at launch
Context Window
- API: 1,000,000 tokens (1M context window)
- Codex: 400,000 tokens
This massive context window makes GPT-5.5 exceptional for long-document analysis, large codebase reviews, and multi-session research projects.
Real-World Results
OpenAI shared several striking use cases from early testers:
Engineering: Dan Shipper (CEO of Every) described GPT-5.5 as "the first coding model I've used that has serious conceptual clarity" — it successfully diagnosed and proposed a rewrite for a complex post-launch bug that GPT-5.4 could not resolve.
Science: An immunology professor at the Jackson Laboratory used GPT-5.5 Pro to analyze a gene-expression dataset with 62 samples and 28,000 genes, producing a detailed research report he said would have taken his team months.
Business: OpenAI's internal Finance team used Codex powered by GPT-5.5 to review 24,771 K-1 tax forms totaling 71,637 pages — accelerating the task by two weeks compared to the prior year.
Mathematics: GPT-5.5 helped discover a new proof about Ramsey numbers — a landmark result in combinatorics — later verified in the Lean proof assistant.
Safety and Preparedness
OpenAI classified GPT-5.5's cybersecurity and biological/chemical capabilities as "High" under its Preparedness Framework. The company deployed stricter classifiers for cyber risk and introduced a Trusted Access program for verified defenders working on critical infrastructure. GPT-5.5 did not reach "Critical" cyber capability level.
How to Access GPT-5.5
- ChatGPT: Select GPT-5.5 or GPT-5.5 Pro from the model picker (Plus/Pro/Business/Enterprise)
- Codex: Available to Plus, Pro, Business, Enterprise, Edu, and Go plans
- API: Use model strings
gpt-5.5orgpt-5.5-proin the Responses or Chat Completions APIs
Platforms like Framia.pro integrate the latest OpenAI models including GPT-5.5, giving teams ready-to-use AI workflows for coding, research, and business automation without requiring direct API setup.
Summary
GPT-5.5 is OpenAI's most capable and production-ready model to date. Its combination of top-tier coding performance, 1M-token context window, improved knowledge work capabilities, and groundbreaking scientific research potential makes it a significant step forward — delivered without sacrificing inference speed. Whether you're a developer, researcher, or enterprise team, GPT-5.5 is the model to build on in 2026.