Baidu, one of China’s tech giants, is shaking up the AI scene with the launch of two cutting-edge models. They’re confident these models surpass those from DeepSeek and OpenAI, at least according to certain benchmarks. This move ramps up the competition in the large language model (LLM) arena.
Let’s dive into what Baidu has unveiled: Ernie 4.5, a versatile model that handles images, audio, and video, and Ernie X1, their first model geared towards multimodal reasoning. You can check them out for free on Baidu’s website. According to Baidu, Ernie 4.5 outperforms OpenAI’s GPT-4o on several benchmarking platforms like CCBench and OCRBench, as shared on their WeChat channel.
When it comes to text processing, Ernie 4.5 has reportedly outdone DeepSeek V3 and is neck and neck with OpenAI’s GPT-4.5. Baidu was the first to introduce an LLM in China back in March 2023, shortly after OpenAI’s ChatGPT took the world by storm. However, the landscape has changed as other Chinese companies have joined the race over the past two years.
Baidu’s latest models are part of their strategy to maintain a strong foothold in China’s bustling AI market. While DeepSeek is known for its open-source strategy, other big names like Alibaba, Tencent, and Bytedance are quickly gaining traction with their own models.
Though Baidu hasn’t shared specific benchmark results for Ernie X1, they claim it delivers similar performance to DeepSeek R1 at half the cost. If you’re considering integrating Ernie X1 into your business, access to its API is priced at 2 yuan (about $0.28) per million token inputs and 8 yuan per million token outputs.
