gpt-5.4-nano

gpt-5.4-nano is the lowest-cost model in the OpenAI GPT-5.4 series for simple high-frequency tasks. The official model catalog recommends using the nano level for throughput-prioritized scenarios, such as label classification, text cleaning, short summaries, field extraction, routing judgment, batch quality inspection, and lightweight customer service intent recognition. It is suitable for completing pre-screening or post-formatting in a multi-model architecture, rather than undertaking complex reasoning and open research. When used, you should limit the output to an explicit schema, prepare evaluation samples that cover edge cases, and refer to human or higher-capability models when confidence is low or rules conflict.

Context

400000 tokens

Max output

128000 tokens

Input

llm image

Output

llm