OpenAI released OpenAI o1-mini, a reasoning model designed to excel at Math and coding. OpenAI o1-mini highlights how models with focus is the next frontier.
The company also said o1-mini's cost is 80% cheaper than OpenAI 01-preview. ChatGPT Plus, Team, Enterprise, and education users can use o1-mini as an alternative to o1-preview to save money.
OpenAI said it trained o1 models without broad knowledge and optimized for STEM reasoning.
In a high school AIME math competition, o1-mini at 70% is competitive with o1 at 74.4%, but with lower costs. Coding had comparison. In other words, enterprises will determine the accuracy needed and optimize costs. The new models from OpenAI deliver much better math performance than GPT-4o.
OpenAI also made the case that slower answers are better for STEM subjects. GPT-4o was fast, but incorrect and the o1 models were correct with slower speeds. The company noted:
"Due to its specialization on STEM reasoning capabilities, o1-mini’s factual knowledge on non-STEM topics such as dates, biographies, and trivia is comparable to small LLMs such as GPT-4o mini. We will improve these limitations in future versions, as well as experiment with extending the model to other modalities and specialties outside of STEM."
OpenAI outlined its views on how LLMs will reason going forward, but there appears to be a bit of a shift away from one-do-it-all model. Humans are also beginning to see models more like a toolkit.
The company concluded:
"o1 significantly advances the state-of-the-art in AI reasoning. We plan to release improved versions of this model as we continue iterating. We expect these new reasoning capabilities will improve our ability to align models to human values and principles. We believe o1 – and its successors – will unlock many new use cases for AI in science, coding, math, and related fields."
Recent:
- Anthropic launches Claude Enterprise with 500K context window, GitHub integration, enterprise security
- GenAI may be the new UI for enterprise software
- OpenAI launches GPT-4o fine tuning
- OpenAI and Microsoft: Symbiotic or future frenemies?
- OpenAI's GPT-4o: A look at short-term, mid-term and long-term implications
