In a week when only one young AI company in China upset the whole world of AI, Openai could not remain silent. While the anthropogenic CEO Dario Amodei has found clear words on the facts and development costs of Deepseek, the most precious and well -known IA startup is now the job.
New model features
The new model of AI O3 -Mini is designed to solve tasks in science, mathematics and coding – and it is supposed to be particularly profitable. It replaces the previous O1-Mini and is now available for Chatgpt Plus, Team and Pro users. For the first time, free users can also try OPENAI O3-Mini by selecting “reason” in the composer of messages or by regenerating an answer.
Pricing strategy
Openai has considerably reduced prices compared to its strongest AI models:
Supplier | Model | Item entrance us-hundred pro 1m | Us-hundred pro 1m release of token |
In depth | Deepseek-R1 | 14 | 219 |
OPENAI | O3-min | 110 | 440 |
OPENAI | GPT-4O | 250 | 1000 |
OPENAI | O1-mini | 300 | 1200 |
OPENAI | O1-PREVIEW | 1500 | 6000 |
Anthropic | Claude 3.5 SONNET | 300 | 1500 |
Anthropic | Claude 3 opus | 1500 | 7500 |
Gemini 1.5 pro | 125 | 500 | |
Amazon | Nova Pro | 80 | 320 |
xai | Goer | 500 | 1500 |
Mistral ai | Big | 200 | 600 |
Key-mini key characteristics
- Print call, structured outputs, Developer’s messages support
- Three levels of reasoning (low, medium, high)
- 24% of faster response times (7.7 against 10.16 seconds)
- Immediate access for Chatgpt Plus, Team and Pro users
- Company access after a week
- Availability for the first time for free users
- Increased rate limit: 150 instead of 50 messages per day
- Performance equivalent to O1 in terms of average reasoning
- 39% fewer serious errors
- “Deliberative alignment” for safer responses
- Surpasses the GPT-4O in safety and jailbreak tests
Safety problems
The main drawback of the OPENAI O3-Mini is its increased risk level compared to previous models. The risk of security has been set to “average” in three out of four areas: persuasion, autonomy and chemicals, biological, radiological and nuclear (CBRN substances):
This is important because Openai is already at the limit of what is possible according to their rules. Only models with a “medium” or lower post-attenuation score can be deployed, and only models with a post-attenuation of “high” or lower can be developed. Openai approaches a point where they may not be able to develop or publish new AI models.