GPT-5.5
OpenAI's smartest general-purpose frontier model for professional work
Context window
1,050,000 tokens (128,000 max output)
Input / 1M tokens
$5.00
Output / 1M tokens
$30.00
Provider
OpenAI
Cached input $0.50 per 1M. Prompts with >272K input tokens are billed at 2x input and 1.5x output for the full session (standard, batch, and flex). Batch and Flex are half the standard rate; Priority processing is 2.5x. A gpt-5.5-pro variant is $30 input / $180 output per 1M. · Data verified 2026-07-02
GPT-5.5 is OpenAI's frontier model for complex professional workloads, released April 23, 2026. It builds on GPT-5.4 with stronger reasoning, higher reliability, and improved token efficiency on hard tasks. 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. It has a 1,050,000-token context window with up to 128,000 output tokens, accepts text and image input, and has a knowledge cutoff of December 1, 2025. A gpt-5.5-pro variant trades higher cost for extended reasoning.
Capability index
Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.
How to access it
Generally available in the OpenAI Responses and Chat Completions APIs (model id 'gpt-5.5', snapshot 'gpt-5.5-2026-04-23'), and in ChatGPT. A higher-capability 'gpt-5.5-pro' variant is also available.
Strengths
- ✓State-of-the-art agentic coding and tool use (SWE-bench Verified ~82.6%)
- ✓Very large 1.05M-token context window
- ✓Strong end-to-end task completion across tools (documents, spreadsheets, software operation)
- ✓Improved token efficiency and reliability over GPT-5.4
- ✓Multimodal text + image input
Best for developers who...
When to choose it (and when not to)
Reach for GPT-5.5 when...
- →Complex professional coding, data analysis, and multi-tool agentic workflows
- →Long-document or large-codebase tasks needing a 1M+ context window
- →You want OpenAI's recommended default frontier model for new projects
Look elsewhere if...
- ✕Latency- or cost-sensitive, high-volume workloads (consider GPT-5.5 Instant or GPT-5.4)
- ✕Simple tasks where a cheaper model like GPT-4o would suffice
- ✕Budgets sensitive to the >272K-token long-context surcharge (2x input / 1.5x output)
How to use it
- ›Give it agentic latitude: state the end goal and let it use tools and iterate to completion
- ›Keep prompts under 272K input tokens where possible to avoid the long-context price multiplier
- ›Use cached input for repeated large prefixes to cut costs (cached input is $0.50/1M)
- ›Escalate to gpt-5.5-pro only when a task genuinely needs deeper extended reasoning
Quickstart
Pythonfrom openai import OpenAI
client = OpenAI()
response = client.responses.create(
model="gpt-5.5",
input="Analyze this CSV and write a summary with three charts.",
)
print(response.output_text)Use snapshot 'gpt-5.5-2026-04-23' to pin behavior. gpt-5.5-pro is available for harder reasoning at higher cost.
API model id: gpt-5.5
Benchmarks
| Benchmark | Score | Notes |
|---|---|---|
| SWE-bench Verified | 82.6% | Real-world software engineering; strong agentic coding result for GPT-5.5. |
| SWE-bench Pro | 58.6% | Harder end-to-end coding tasks solved single-pass, per OpenAI. |
Source: OpenAI - Introducing GPT-5.5
Tools powered by GPT-5.5
Compare GPT-5.5
GPT-5.5 vs Claude Sonnet 5
Anthropic - 1M ctx
GPT-5.5 vs GPT-5
OpenAI - 400,000 tokens (128,000 max output) ctx
Compare GPT-5.5 with any other model
Build a comparison →All model comparisons →