DataLand ^ ChatGPT: To Start a New AI-oriented Social Paradigm (3)

GPT-4 is a large multimodal model that can accept image and text inputs while emitting text outputs. Experiments show that GPT-4 exhibits human-level performance on various professional and academic benchmarks. For instance, it successfully completes a mock bar exam with a score in the top 10% of test takers, as opposed to GPT-3.5, whose score was in the bottom 10%.

Over the course of six months, OpenAI iteratively aligned GPT-4 using lessons from adversarial testing programs as well as ChatGPT, yielding the best-ever results on factuality and steerability.

Let’s try GPT-4 with the following picture, what’s its meme about?

From this dialogue, we can see GPT-4’s potent picture recognition capabilities and it resembles another stalk, whose tone is reminiscent of Spock (one of Star Trek’s main characters).

After the release of GPT-4, OpenAI directly upgraded ChatGPT. ChatGPT Plus subscribers will get GPT-4 access on chat.openai.com with a usage cap.

To access the GPT-4 API (which uses the same ChatCompletions API as gpt-3.5-turbo), users can sign up for the waitlist. OpenAI will invite some developers to experience.

Developers can send text-only requests to the gpt-4 model once they have access to the API (image inputs are still in limited alpha). Pricing is $0.03 per 1k prompt tokens and $0.06 per 1k completion tokens. Default rate limits are 40k tokens per minute and 200 requests per minute.

GPT-4 has a context length of 8,192 tokens. OpenAI also provides limited access to 32,768–context (about 50 pages of text) version, which will also be updated automatically over time (current version gpt-4–32k-0314, also supported until June 14). Pricing is $0.06 per 1K prompt tokens and $0.12 per 1k completion tokens.

In any case, the picture input function will be enhanced by GPT-4, which will strongly encourage the competitors on the racetrack of AIGC storage — in which SINSO has been playing a strong role. The Dataland&ChatGPT version that SINSO will announce on March 23 will employ SINSO tokens in place of payments, so you won’t even need to pay to use the service — just stake SINSO tokens. Of course, SINSO tokens can also be used to purchase the recharge consumable package, which is less expensive than openAI’s estimate.

GPT-4 is still in the alpha phase, and the GPT-3.5 version will be available in the next few days, let’s wait and see.