What happened
OpenAI released GPT-5.4 mini and nano, its most capable small models for high-volume workloads. GPT-5.4 mini significantly improves over GPT-5 mini, running over 2x faster and approaching GPT-5.4 performance on SWE-Bench Pro (54.4%) and OSWorld-Verified (72.1%). GPT-5.4 nano, the smallest and cheapest, targets classification and coding subagents. Mini features a 400k context window, costing $0.75 per 1M input and $4.50 per 1M output tokens; nano costs $0.20 per 1M input and $1.25 per 1M output tokens.
Why it matters
These models enable composable AI architectures, allowing larger models to plan while smaller, faster, and cheaper subagents execute specific tasks. Platform engineers can now optimise workflows by delegating less reasoning-intensive work to GPT-5.4 mini, which uses 30% of the GPT-5.4 quota in Codex, reducing operational costs and improving responsiveness for coding assistants and multimodal applications. This shifts system design towards specialised model deployment over monolithic solutions, offering a new cost-performance frontier for high-volume AI applications.
Subscribe for Weekly Updates
Stay ahead with our weekly AI and tech briefings, delivered every Tuesday.
