GPT-5.4 mini and GPT-5.4 nano, two new small-scale artificial intelligence models from OpenAI, became available on March 9, 2026, targeting high-volume, latency-sensitive workloads. The company said both models bring capabilities from the larger GPT-5.4 to faster and more cost-efficient deployments.
GPT-5.4 mini Performance and Availability
GPT-5.4 mini runs more than 2x faster than GPT-5 mini, according to OpenAI. The model approaches GPT-5.4-level performance on several evaluations, including SWE-Bench Pro and OSWorld-Verified. On OSWorld-Verified, it scored 72.1% accuracy, compared to 75.0% for the full GPT-5.4 model and 39.0% for GPT-5 mini.
Pricing and API Specifications
In the API, GPT-5.4 mini supports text and image inputs, tool use, function calling, web search, file search, computer use, and skills. It carries a 400,000-token context window and costs $0.75 per 1 million input tokens and $4.50 per 1 million output tokens. GPT-5.4 nano, meanwhile, costs $0.20 per 1 million input tokens and $1.25 per 1 million output tokens, and is available exclusively through the API.
Coding and Multimodal Use Cases
OpenAI said GPT-5.4 mini is designed for coding workflows that require fast iteration, including targeted edits, codebase navigation, front-end generation, and debugging loops. The model also handles multimodal tasks, specifically interpreting screenshots of dense user interfaces for computer-use applications. In benchmark testing, GPT-5.4 mini consistently outperformed GPT-5 mini at similar latencies while approaching GPT-5.4-level pass rates.
Aabhas Sharma, CTO at Hebbia, said the model delivered strong results in their evaluations.
“GPT-5.4 mini delivers strong end-to-end performance for a model in this class. In our evaluations it matched or exceeded competitive models on several output tasks and citation recall at a much lower cost. It also achieved higher end-to-end pass rates and stronger source attribution than the larger GPT-5.4 model.”
Aabhas Sharma, CTO at Hebbia
Deployment Across ChatGPT and Codex
In Codex, GPT-5.4 mini is available across the Codex app, CLI, IDE extension, and web interface. It consumes only 30% of the GPT-5.4 quota, allowing developers to handle simpler coding tasks at roughly one-third the cost. Codex can also delegate tasks to GPT-5.4 mini subagents, so less reasoning-intensive work runs on the cheaper model. In ChatGPT, GPT-5.4 mini is available to Free and Go users via the “Thinking” feature in the + menu. For all other users, it serves as a rate-limit fallback for GPT-5.4 Thinking.
OpenAI said the release reflects a broader design pattern in which larger models handle planning and coordination while smaller models execute narrower subtasks in parallel. The company noted that GPT-5.4 nano is recommended specifically for classification, data extraction, ranking, and coding subagents that handle simpler supporting tasks. Furthermore, OpenAI said details on model safeguards are available in the System Card addendum on its Deployment Safety Hub.