- OpenAI’s GPT-5.5 launches at $30 per million output tokens—double the previous tier’s cost.
- The model matches GPT-5.4’s speed while promising ‘a new class of intelligence’ for agent workflows.
- Codenamed ‘Spud’ during development, it’s now available to ChatGPT Plus and API users.
OpenAI released GPT-5.5 on Thursday, a new AI model that the company claims matches the latency of GPT-5.4 while delivering what it calls ‘a new class of intelligence’—at prices that effectively double API costs for developers.
The model, codenamed ‘Spud’ during development, is available to ChatGPT Plus subscribers and through OpenAI’s API at $5 per million input tokens and $30 per million output tokens, representing a significant pricing adjustment from previous tiers. OpenAI announced the launch in a blog post Thursday afternoon.
‘GPT-5.5 is our most capable model yet,’ OpenAI wrote. ‘It matches GPT-5.4’s latency while delivering higher intelligence, and it’s designed for real-world agent workflows.’ The company emphasized the model’s improved performance on coding benchmarks and complex reasoning tasks.
GPT-5.5 Pricing and Performance
The pricing structure places GPT-5.5 between GPT-5.4 and the higher-end GPT-5.5 Turbo in OpenAI’s lineup. At $30 per million output tokens, developers pay roughly twice what GPT-5.4 costs for the latency-matched throughput. The model processes context windows up to 256,000 tokens, the same as its predecessors.
Early testers reported improvements in code generation quality and long-form content coherence, though the price jump may limit adoption for cost-sensitive applications. OpenAI positioned the release as targeting ‘mission-critical workloads’ where accuracy matters more than per-token economics.
GPT-5.5 also introduces improved tool use capabilities, enabling better integration with external APIs and databases—features critical for the autonomous agent workflows OpenAI has been pushing. The model supports 40 languages and handles both text and vision inputs.
OpenAI’s move follows Anthropic’s recent Claude 4 launch and comes amid intensifying competition in the frontier model space. Amazon’s Bedrock and Google Cloud have also been expanding their model offerings to compete with OpenAI’s API.
