OpenAI releases "Spud" GPT-5.5 model
The new model processes 1 million tokens and costs 60% less than GPT-4o.
OpenAI has launched GPT-5.5, internally codenamed 'Spud', marking a significant leap in performance and efficiency. The model supports a 1 million token context window, enabling processing of entire codebases, lengthy legal documents, or multi-hour meeting transcripts in a single pass. Inference speed is 3x faster than GPT-4o, with a 60% reduction in cost per token, making it accessible for high-volume enterprise applications.
Spud introduces native function calling for AI agents, allowing the model to autonomously execute API calls and workflows without external orchestration. Early benchmarks show a 15% improvement on GPQA (graduate-level Q&A) and 20% on HumanEval (code generation). The model is available via OpenAI's API and ChatGPT Plus, with a new 'Spud Pro' tier offering priority access for $200/month. Analysts note Spud's efficiency could disrupt cloud AI costs, potentially saving enterprises 40-50% on inference budgets.
- 1M token context window for processing entire codebases or long documents
- 3x faster inference and 60% lower cost per token vs GPT-4o
- Native function calling enables autonomous AI agent workflows
Why It Matters
Spud slashes AI costs by 60% while enabling complex agentic workflows, reshaping enterprise deployment economics.