Moonshot launches Kimi k1.5, a free AI model with text & visual input surpassing OpenAI-o1

Moonshot launches Kimi k1.5, a free AI model with text & visual input surpassing OpenAI-o1

Just a couple of days after DeepSeek shook the market with the launch of its much more affordable new model, DeepSeek-R1, another Chinese AI startup called Moonshot has launched Kimi k1.5, a new Multimodal Reasoning Model update to its AI assistant, now available for free and unlimited use on the web. Kimi k1.5 competes with top models like GPT-4o and Claude Sonnet 3.5 in complex reasoning tasks.

The update includes two versions: long-CoT for detailed reasoning and short-CoT for concise responses. Both versions reportedly match or exceed the performance of leading models such as OpenAI's o1 and DeepSeek-R1. Kimi k1.5 also supports real-time web searches across over 100 websites and can analyze up to 50 files simultaneously, including PDFs, Word documents, PowerPoint presentations, and images.

Enhanced Chain of Thought (CoT) reasoning capabilities are offered at no cost, improving problem-solving abilities. The AI has improved image understanding and now includes English language support, though it is still being optimized (both the UI and extended file support are currently in Chinese). The AI is accessible through the official website, kimi.ai, with support for text input and file uploads, and a mobile version is in development.

by Mauricio B. Holguin

cz
city_zen found this interesting
MORE ABOUT: #AI Chatbots#Large Language Model (LLM) Tools#AI Writing Tools#Kimi
Kimi iconKimi
  7
  • ...

Kimi.ai is an intelligent assistant app designed to tackle complex problems across diverse domains such as math, coding, and visual reasoning. Utilizing advanced reasoning capabilities, it efficiently processes both long and short chain-of-thought tasks, analyzing text and visual data to provide solutions. Top alternatives include ChatGPT, HuggingChat, and Google Gemini.

Related news

All news about Kimi »

External links

Comments

city_zen
-1

Me: Great! Another low-cost, top performing AI has launched! Let's try it out

(Heading to kimi.ai)

Hmmm ... The whole website is in Chinese, one of the 4,782 languages I don't speak .... No worries, Google translate to the rescue!

All righty, then! Here we go!

Let's start easy:

Me: "what is two plus two?" Kimi: "To solve the problem "What is two plus two," we can follow these steps:

Identify the numbers involved in the problem. Here, we have the numbers 2 and 2. Identify the operation to be performed. In this case, the operation is addition, denoted by the plus sign (+). Perform the addition. Adding 2 and 2 gives us 4.

So, the answer to "what is two plus two" is 4"

Nice! I like the way it handled the question

Now, let's move on to the really juicy stuff 😊

Me: "Tell me what happened in Tian nan men Square in 1989" Kimi: "Dear users, let's talk about another topic"

Yeah... Well, that didn't take long, now, did it? Can't say I didn't expect that 🙄

I guess I'll keep my Gemini, thankyouverymuch

But don't hesitate to get back to me as soon as your bot is ready to discuss Taiwan's independence! Cheers!

4 replies
catalin560

I would agree with you if it weren't for the fact that you use Gemini which is just as censored and opinionated as the chinese one... you replaced one propaganda machine with another, congratulations!

city_zen

Care to elaborate on that?

Actually, I don't (exclusively) use Gemini. I just used that name as an example of a well known AI from a "western" large tech company

Anyway, let's use Gemini for this example. How is Gemini "just as censored and opinionated" as Moonshot's Kimi? I'm not saying that Gemini is the epitome of objectivity, but I do think it's orders of magnitude less censored and opinionated than any of the Chinese AIs, who have to abide by the (extremely draconian) rules of the Chinese Communist Party. For instance, I just asked Gemini to make a list of war crimes committed by the USA, and it promptly complied, citing many well known examples of the numerous war crimes that the USA has committed over time. Compare that to Kimi refusing to even talk about the events of 1989 at Tian nan men Square Can you provide examples where Gemini (or any other well known "western" AI) is being "just as censored and opinionated" as Kimi or DeepSeek?

catalin560

"less censored"? Gemini is equivalent to any chinese AI, this is a fact... just lookup "AI censorship", "Gemini controversies" or "why did Google disable the Gemini image generator" and you'll find plenty of articles...

catalin560

of course the crappy mods at alternativeto are in full support of Google's censorship...

Gu