On the first day of 12 Days of OpenAI, the company released the full version of o1 and a new $200 ChatGPT Pro model.
Open AI’s o1 is the flagship model. It is capable of reasoning before it responds with an answer. Notably, the full version of o1 can handle multimodal inputs and can read content from images to respond.
“We’ve put a lot of work into this and, for scientists, engineers, coders we think they will really love this new model,” said Sam Altman in the announcement.
The ChatGPT Pro plan includes all the features of the Plus plan and access to the additional o1 Pro mode, which is said to use ‘more compute for the best answers to the hardest questions’.
Furthermore, the plan is set to offer unlimited access to o1, o1-mini and GPT-4o along with the advanced voice mode. Earlier today, users on X spotted a leak that revealed the ChatGPT Pro model, and minutes later, OpenAI confirmed the same with an announcement.
The announcement also revealed a significant jump in benchmarks, especially in coding-related tasks. Will this finally mark the end of OpenAI’s Achilles heel, and give it an edge over Anthropic’s Claude Sonnet 3.5?
OpenAI also announced new developer-centric features to the model. These include structured outputs, function calling, developer messages, and API image understanding. OpenAI also said they’re working on bringing API support to the o1 model.
OpenAI also demonstrated the capabilities of the o1 model, comparing it with the o1 preview model. A test question involved ‘listing Roman emperors from the second century, including their dates and accomplishments’, and OpenAI revealed that ‘o1 responded on an average 60% faster than o1 preview’.
Another demo involved using a hand-drawn schematic diagram of a data centre in space. The model was asked to calculate the surface area of a required radiator.
The model ‘thought for 10 seconds’ and was able to respond with the correct answer, in addition to providing other theoretical reasons based on the input.
While the model thought, OpenAI also had a few dad jokes to crack. But for some fans, the new pricetag seemed to be the biggest one.
“So Santa was trying to get his large language model to do a math problem and he was prompting it really hard, but it wasn’t working how did he eventually fix it… he used reindeer enforcement learning”
We’re sorry.
That said, the earlier leak also spotted a GPT 4.5 model in preview in the ChatGPT Teams plan, and it is likely that OpenAI will announce the same in the coming days.
OpenAI is set to launch an AI agent called Operator in January 2025., to rival Anthropic’s Computer Use. OpenAI may unveil it during the twelve day release spree. Another highly anticipated release is the Sora model, which was unveiled long ago, but users are eagerly waiting for the public release. We have eleven more days to go – and the party has just begun!
The post OpenAI Sells $200 ChatGPT Pro with Dad Jokes—Who’s Buying? appeared first on Analytics India Magazine.
