Chinese AI company MiniMax is releasing new models that it claims can compete with the best in the industry

Chinese firms continue to release AI models that rival the capabilities of systems developed by OpenAI and other US-based AI companies.

this week, MiniMaxA startup backed by Alibaba and Tencent raised Nearly $850 million in venture capital and over $2.5 billion in valuation, made his debut three new models: MiniMax-Text-01, MiniMax-VL-01 and T2A-01-HD. MiniMax-Text-01 is a text-only model, while MiniMax-VL-01 can understand both images and text. The T2A-01-HD, meanwhile, produces audio — especially speech.

MiniMax claims that MiniMax-Text-01, which has a size of 456 billion parameters, outperforms Google’s recently demonstrated models. Gemini 2.0 Flash on benchmarks such as MATH and SimpleQA, which measure a model’s ability to answer math problems and fact-based questions. The parameters roughly match the problem-solving abilities of the model, and models with more parameters generally perform better than models with fewer parameters.

As for the MiniMax-VL-01, MiniMax says it’s a competitor to Anthropic. Claude 3.5 Sonnet In assessments that require multimodal understanding, such as ChartQA, which instructs models to respond to graph and chart-related queries (eg, “What is the peak value of the orange line on this graph?”). Admittedly, the MiniMax-VL-01 is no better than the Gemini 2.0 Flash in many of these tests. OpenAIs GPT-4o and Meta Llama 3.1 he also beat him several times.

Note that MiniMax-Text-01 has a rather large context window. A model’s context, or context window, refers to the input (such as text) that the model considers before generating output (additional text). With a context window of 4 million tokens, MiniMax-Text-01 can analyze about 3 million words at a time, or five copies of War and Peace.

For context (no pun intended), MiniMax-Text-01’s context window is about 31 times larger than GPT-4o and Llama 3.1.

MiniMax’s latest model, released this week, is the T2A-01-HD speech-optimized audio generator. The T2A-01-HD can generate a synthetic voice with adjustable cadence, tone and tenor in about 17 different languages, including English and Chinese, and clone the voice from just 10 seconds of recording.

MiniMax has not published benchmark results comparing the T2A-01-HD to other sound-generating models. But to this reporter’s ears, the T2A-01-HD’s outputs sound on par with audio models. Meta and startups like PlayAI.

With the exception of the T2A-01-HD, which is only available through MiniMax’s API and the Hailuo AI platform, MiniMax’s new models can be downloaded from GitHub and the Hugging Face AI development platform.

The fact that models are “open” does not mean that they are not closed in certain aspects. MiniMax-Text-01 and MiniMax-VL-01 not really open source in the sense that MiniMax has omitted the components (such as training data) needed to recreate them from scratch. Moreover, they are under MiniMax’s restrictive license, which prohibits developers from using the models to improve competing AI models, and requires platforms with more than 100 million monthly active users to request a special license from MiniMax.

MiniMax was founded in 2021 by former employees of SenseTime, one of the largest artificial intelligence firms in China. The company’s projects include applications such as Talkie, an artificial intelligence-powered role-playing platform. Character AIand MiniMax’s text-to-video models released on Hailuo.

Some of MiniMax’s products have caused minor controversy.

Talkie, which was pulled from Apple’s App Store in December for unspecified “technical” reasons, features artificial intelligence avatars of public figures such as Donald Trump, Taylor Swift, Elon Musk and LeBron James. program.

Broadcast magazine in December informed The fact that the MiniMax’s video generators can reproduce the logos of British TV channels suggests that the MiniMax models are trained on the content of those channels. MiniMax is reported is brought to court by iQIYI, a Chinese video streaming service that claims MiniMax illegally trained on iQIYI’s copyrighted recordings.

The new MiniMax models come just days after the end of the Biden Administration he suggested Tighter export rules for Chinese enterprises and restrictions on AI technologies. Companies in China are already barred from buying advanced AI chips, but if the new rules go into effect as written, companies will face tougher caps on both the semiconductor technology and the models needed to power sophisticated AI systems.

On Wednesday, the Biden Administration announced additional measures are aimed at keeping sophisticated chips out of China. Chip foundries and packaging companies looking to export certain chips will be subject to broader licensing requirements unless they do more checks and due diligence to prevent their products from reaching Chinese customers.