Baidu Announces 'ERNIE 4.5' and 'ERNIE X1' Multimodal AI Models Comparable to DeepSeek-R1
Chinese tech company Baidu announced the multimodal AI models ' ERNIE 4.5 ' and ' ERNIE X1 ' on March 16, 2025. According to Baidu, ERNIE X1 will provide nearly equivalent performance to DeepSeek-R1 at half the cost.
We've just unveiled ERNIE 4.5 & X1! ????
pic.twitter.com/cLKVHYvbzw — Baidu Inc. (@Baidu_Inc) March 16, 2025
As a deep-thinking reasoning model with multimodal capabilities, ERNIE X1 delivers performance on par with DeepSeek R1 at only half the price. Meanwhile, ERNIE 4.5 is our latest foundation model and new-generation native multimodal model.…
Baidu Unveils ERNIE 4.5 and Reasoning Model ERNIE X1, Makes ERNIE Bot Free Ahead of Schedule
https://www.prnewswire.com/news-releases/baidu-unveils-ernie-4-5-and-reasoning-model-ernie-x1--makes-ernie-bot-free-ahead-of-schedule-302402490.html
Introducing ERNIE 4.5 | Our New-Generation Native Multimodal Model - YouTube
Baidu said, 'ERNIE 4.5 realizes collaborative optimization through joint modeling of multiple modalities, demonstrating comprehensive improvements in understanding, generation, inference, and memory, as well as notable improvements in hallucination prevention, logical reasoning, and coding capabilities.' It also posted a graph comparing the benchmark results of ERNIE 4.5 with those of GPT-4o, DeepSeek-V3, and GPT-4.5.
ERNIE 4.5 achieves collaborative optimization through joint modeling of multiple modalities, exhibiting comprehensive improvements in understanding, generation, reasoning and memory, along with notable enhancements in hallucination prevention, logical reasoning, and coding… pic.twitter.com/8Iu3pmpMtV
— Baidu Inc. (@Baidu_Inc) March 16, 2025
ERNIE 4.5 is a multimodal model that can understand not only text but also images, audio, and video. It also has strong intelligence and contextual awareness, and can easily understand internet memes and cartoons.
And the inference-specialized ERNIE X1 is Baidu's first 'tool-enabled multimodal deep thinking inference model,' and excels in Q&A on Chinese knowledge, literary creation, manuscript writing, dialogue, logical inference, and complex calculations. Baidu also claims that ERNIE X1 can provide the same performance as DeepSeek-R1 at half the cost.
The conversation with ERNIE 4.5 can be accessed free of charge by registered Baidu individual users from the following page:
A word from the heart
https://yiyan.baidu.com/
The APIs of ERNIE 4.5 and ERNIE X1 can be accessed directly through Baidu AI Cloud's Qianfan Foundation Model Platform. The usage fees for corporations are set quite low, with ERNIE 4.5 starting at 0.004 RMB (approx. 0.082 yen) per 1,000 tokens for input and 0.016 RMB (approx. 0.33 yen) per 1,000 tokens for output. ERNIE X1 is 0.002 RMB (approx. 0.041 yen) per 1,000 tokens for input and 0.008 RMB (approx. 0.16 yen) per 1,000 tokens for output.
'2025 will be a critical year for the development and iteration of large-scale language models and technologies. With the release of ERNIE 4.5 and ERNIE X1, Baidu will continue to invest in artificial intelligence, data centers and cloud infrastructure to improve its AI capabilities and develop smarter, more powerful next-generation models,' Baidu said.
Related Posts:
