OpenAI released GPT-5.5 on April 23, its newest frontier model and the company’s most capable agentic system to date, designed to accept a messy multi-part task and plan, execute, and complete it without requiring step-by-step human guidance.
Summary
- OpenAI released GPT-5.5 on April 23, 2026, positioning it as a new class of agentic intelligence built to handle autonomous, multi-step computer work with minimal instruction.
- The model scores 82.7% on Terminal-Bench 2.0, 84.9% on GDPval across 44 knowledge work occupations, and 78.7% on OSWorld-Verified, which tests autonomous operation of real computer environments.
- GPT-5.5 replaces GPT-5.4 as the default model for ChatGPT Plus, Pro, Business, and Enterprise users, with a Pro variant available for higher-accuracy, longer-horizon tasks.
OpenAI released GPT-5.5 on April 23, billing it as a “new class of intelligence for real work” that marks a deliberate shift away from chatbot positioning toward autonomous agentic execution. The model is now available for ChatGPT Plus, Pro, Business, and Enterprise users, with a GPT-5.5 Pro variant available for heavier workloads, just six weeks after GPT-5.4 launched.
OpenAI GPT-5.5 Agentic Model Reframes AI as a Worker, Not a Chatbot
The core pitch for GPT-5.5 is capability without hand-holding. OpenAI described the model as being able to “look at an unclear problem and figure out what needs to happen next,” with Greg Brockman, OpenAI co-founder and president, saying the model is “way more intuitive to use” and can carry significantly more of the work itself compared to its predecessor. On Terminal-Bench 2.0, a benchmark testing complex command-line tasks requiring planning, iteration, and tool coordination, GPT-5.5 scores 82.7%, ahead of Claude Opus 4.7 at 69.4%. On GDPval, which measures agent performance across 44 occupations of knowledge work, it scores 84.9%. On OSWorld-Verified, a benchmark testing whether a model can autonomously operate real computer environments, it reaches 78.7%. OpenAI Chief Research Officer Amelia Glaese said GPT-5.5 is “definitely our strongest model yet on coding, both measured by benchmarks and based on the feedback that we’ve gotten from trusted partners.” The model also runs at the same per-token latency as GPT-5.4 while using significantly fewer tokens to complete the same tasks.
A Deliberate Positioning Shift From Chat to Execution
OpenAI is no longer selling a chat completion API with GPT-5.5. The language used across the launch is uniformly about outcomes rather than quality metrics. The company describes GPT-5.5 as capable of handling coding, online research, data analysis, document creation, spreadsheet work, and software operation as a continuous workflow rather than a series of isolated prompts. API pricing reflects the same message: standard GPT-5.5 is priced at $5 per million input tokens and $30 per million output, while GPT-5.5 Pro is $30 per million input and $180 per million output, pricing designed for agent deployments where long context and high token generation per task are standard requirements. As crypto.news reported, OpenAI has been systematically expanding into professional workflow infrastructure, having launched a financial-services AI stack in March 2026 that connects ChatGPT to FactSet, Third Bridge, Excel, and Google Sheets. GPT-5.5 is the model that makes those integrations genuinely autonomous rather than merely interactive.
What GPT-5.5 Means for the Competitive Landscape
The launch arrives as OpenAI competes directly with Anthropic’s Claude Mythos Preview, a cybersecurity-focused model that scored 82.0% on Terminal-Bench 2.0 compared to GPT-5.5’s 82.7%. OpenAI was explicit that GPT-5.5 does not exceed its Critical cybersecurity risk threshold but meets the High risk classification, requiring enhanced safeguards against misuse in bio and cyber contexts. As crypto.news documented, OpenAI is projecting $14 billion in losses in 2026 while targeting $174 billion in revenue by 2030, making the shift toward agentic AI and enterprise workflow ownership the fastest available path to the kind of sustained revenue that would make those projections credible. As crypto.news tracked, the company crossed $10 billion in annual recurring revenue in mid-2025 and is expected to approach $30 billion in 2026 as agentic products capture a larger share of the revenue mix.
Sam Altman posted on X the day of the launch that OpenAI “believes in iterative deployment” and described GPT-5.5 as already a smart model with rapid improvements expected, framing the release as a foundation rather than a ceiling.
