Model positioning
GPT-5 mini is a faster, lower-cost version of GPT-5, suited to cost-sensitive, low-latency, and high-throughput scenarios, particularly workloads with well-defined tasks and precise prompts. (developers.openai.com)
Capabilities and uses
- Supports text input and output, image input, and reasoning tokens; can be used in applications for text processing, visual understanding, programming, and tool calling. (developers.openai.com)
- Supports streaming, function calling, and structured output; parameters can control reasoning effort and response detail, and tools such as search are supported. (developers.openai.com)
- The context window is 400,000 tokens, and the maximum output is 128,000 tokens; the knowledge cutoff date is May 31, 2024. (developers.openai.com)
gpt-5-mini-2025-08-07is a snapshot ID used to lock in a specific model version. (developers.openai.com)
Limitations and status
Audio or video input and output are not supported, nor is fine-tuning. The model may make mistakes; results should be verified for high-risk uses. (developers.openai.com)
As of November 18, 2026, the model and its snapshot had been officially marked as deprecated and were scheduled to shut down on December 11, 2026; this does not mean they had already shut down on the query date. (developers.openai.com)
API pricing
The standard text token prices published by OpenAI are in US dollars (USD): $0.25 per 1 million input tokens and $2.00 per 1 million output tokens; cached input eligible for prompt caching costs $0.025 per 1 million tokens. Tool calls may incur additional charges; prices for other billing methods, such as batch processing, are not included in the standard prices above. Price source: Official GPT-5 mini model documentation; query date: November 18, 2026. (openai.com)
Release and official sources
- Release date: August 7, 2025. (openai.com)
- GPT-5 launch announcement for developers
- GPT-5 mini model documentation
- Model deprecation schedule