o4-mini
Lightweight reasoning + full tool use + native multimodality; free-tier; strong AIME/GPQA.
Latest lightweight reasoning model released to all ChatGPT users combining efficiency with advanced capabilities. First mini model supporting full tool use and native multimodality for text and image processing. Released April 16, 2025 to free-tier and paid users. Supports chain-of-thought reasoning for transparent problem-solving. o4-mini-high variant available to paid subscribers offering enhanced reasoning effort. Achieves 92.7% accuracy on AIME 2025 mathematics examination. Scores 81.4 on GPQA Diamond for Ph.D.-level science questions. Delivers substantial improvement in chemistry tasks compared to earlier models. Combines lightweight o4-mini architecture with tool support eliminating previous mini model limitations.
Reviews
No Reviews Yet
Be the first to share your experience with this AI tool
More models from ChatGPT
Enterprise GPT-5.2 tier using parallel test-time compute for highest accuracy; slower but best.
Cost-efficient GPT-5 variant: 400K context, faster inference, adjustable reasoning, ~2x cheaper.
Lightweight GPT-5: 400K context, minimal latency, adaptive reasoning, cheapest for throughput.
Compact GPT-4o: strong reasoning/coding for cost, 128K context, fast for real-time apps.