o1 Preview is an LLM designed for complex reasoning tasks that require multiple steps to solve. It excels in spending more time thinking before responding, solving harder problems, and demonstrating improved capabilities in advanced math, science, and coding challenges.

Some other noteworthy features of o1 Preview include its enhanced resistance to jailbreaks and adversarial prompts through the use of chain-of-thought rationales, as well as its ability to navigate complex challenges without fixating on failed reasoning paths.

Metric	Value
Parameter Count	Unknown
Mixture of Experts	Unknown
Context Length	128,000 tokens
Multilingual	Unknown
Quantized*	Unknown

*Quantization is specific to the inference provider and the model may be offered with different quantization levels by other providers.