← Models

GPT-5.5

OpenAI · GPT-5

Still-available GPT-5.5 API route for pinned agentic coding, professional work, research, and migration baselines.

Part of GPT family · Other versions: GPT-5.6 Luna , GPT-5.6 Terra , GPT-5.6 Sol , GPT-5.5 Pro , GPT-Rosalind , GPT-5.4 mini , GPT-5.4 nano , GPT-5.4 Pro , GPT-5.4 , GPT-5.3 , GPT-5.3-Codex , GPT-5.2-Codex , GPT-5.2 Pro , GPT-5.2 , GPT-5-Codex , GPT-5 mini , GPT-5 nano , GPT-5 , GPT-4.1 , GPT-4o mini Transcribe , GPT-4o mini TTS , GPT-4o Transcribe , Computer Use Preview , GPT-4o mini , GPT-4o
Type
language
Context
1M tokens
Max Output
128K tokens
Status
current
Input
$5/1M tok
Output
$30/1M tok
API Access
Yes
License
proprietary
reasoning coding agentic tool-use computer-use long-context professional-workflows
Released April 2026 · Updated July 10, 2026

Overview

Freshness note: Model capabilities, limits, and availability can change quickly. This profile is a point-in-time snapshot last verified on July 10, 2026.

GPT-5.5 is OpenAI’s previous frontier GPT-5 model for complex real-world work. OpenAI launched it on April 23, 2026, rolled it through ChatGPT and Codex, and still publishes gpt-5.5 as a non-deprecated API model. The July 9 general availability of GPT-5.6 Sol, Terra, and Luna changes its role: GPT-5.5 is now most useful as a pinned production route, regression baseline, or deliberate migration source rather than the first model to evaluate for a new build.

The underlying capability remains relevant for agentic coding, professional analysis, research, and computer use. Existing deployments do not need to move blindly, but new deployments should compare GPT-5.6 on representative tasks before standardizing here.

Capabilities

GPT-5.5 is strongest in agentic coding, computer use, knowledge work, and early scientific research workflows. OpenAI highlights better persistence across long tasks, stronger tool use, fewer retries, and higher token efficiency on Codex tasks than GPT-5.4.

The practical upgrade is not just benchmark score. It is better behavior across the full work loop: understand the goal, inspect context, use tools, modify artifacts, test assumptions, and recover when something is ambiguous. That makes it relevant for implementation work, code review, spreadsheet modeling, document-heavy analysis, and source-heavy research.

Technical Details

OpenAI’s API docs list gpt-5.5 with a 1,050,000-token context window, 128,000-token maximum output, text and image input, and text output. In current ChatGPT and Codex product surfaces, exposed limits can still differ by plan, mode, and client. OpenAI’s launch article also says Codex uses a 400K context window.

The current API docs say reasoning.effort supports none, low, medium, high, and xhigh, with medium as the default. GPT-5.5 also keeps the main GPT-5.4 API feature surface, including streaming, function calling, structured outputs, hosted tools, tool search, and computer use through the Responses API.

The public system card also frames GPT-5.5 as a reasoning model evaluated across safety, hallucination, health, cybersecurity, biology, prompt-injection, and agentic coding risk areas. OpenAI treats GPT-5.5 Pro as the same underlying model with additional parallel test-time compute.

OpenAI’s GPT-5.6 migration guide recommends preserving the current GPT-5.5 reasoning effort as a baseline, then testing the same setting and one level lower. That is more useful than assuming a one-to-one quality or cost shift from headline model pricing alone.

Pricing & Access

GPT-5.5 is live in ChatGPT and Codex for eligible plans, and OpenAI’s API docs now list gpt-5.5 for developer use through the Responses API and Chat Completions.

Published API pricing for gpt-5.5 is:

  • Input: $5.00 per 1M tokens
  • Cached input: $0.50 per 1M tokens
  • Output: $30.00 per 1M tokens

OpenAI also documents two pricing caveats that matter for production planning: prompts above 272K input tokens are priced at 2x input and 1.5x output for the full session, and regional-processing endpoints carry a 10% uplift. Batch and Flex processing remain the lower-cost lanes for supported workloads.

Best Use Cases

Use GPT-5.5 for hard planning, complex repository work, long-running debugging, professional analysis, tool-heavy research, and computer-use workflows where GPT-5.4 is not reliable enough or requires too much cleanup. It is not the economical default for every request. Route high-volume extraction, routine drafting, and narrow classification to smaller or cheaper models.

Comparisons

  • GPT-5.6 Sol (OpenAI): Current flagship generation with the same standard 5/5/30 token price, newer agent features, and stronger official migration priority.
  • GPT-5.6 Terra (OpenAI): Balanced migration target at half GPT-5.5’s standard token price.
  • GPT-5.4 (OpenAI): Still relevant for compatibility, cost, and teams already standardized on that route, but no longer the main OpenAI frontier recommendation.
  • GPT-5.5 Pro (OpenAI): Higher-compute route for the hardest work and premium API escalation.
  • Claude Opus 4.7 (Anthropic): Strong competing premium model, especially for teams already standardized around Claude products.
  • Gemini 3.1 Pro Preview (Google): Preview-tier alternative with strong multimodal and Google ecosystem integration.