OpenAI: GPT Mini Latest

OpenAI: GPT Mini Latest (~openai/gpt-mini-latest) is a rolling API alias that always resolves to the newest model in OpenAI’s GPT Mini product line, so integrations can track improvements without changing model IDs. On OpenRouter and partner catalogs it is documented with a 400K-token context window, up to 128K output tokens per request, multimodal text-and-image input, tool use, and an August 31, 2025 knowledge cutoff; pricing is typically aligned with the current mini flagship (for example GPT-5.4 mini at $0.75 per million input tokens and $4.50 per million output tokens).

As of early 2026 the alias points at GPT-5.4 mini, OpenAI’s fastest small reasoning model in the GPT-5.4 generation. It is optimized for high-volume coding assistants, subagents, computer-use workflows over screenshots, and low-latency tool orchestration, with configurable reasoning effort up to extra-high on supported endpoints. OpenAI reports more than 2× speed versus GPT-5 mini while closing much of the gap to full GPT-5.4 on professional coding and agent benchmarks.

Benchmark figures below reflect the current GPT Mini family target (GPT-5.4 mini at the highest reported reasoning settings in official OpenAI evals, supplemented by DataLearner leaderboard entries where applicable). Because the alias can change when OpenAI ships a newer mini release, re-validate scores and pricing after any redirect update.