On Monday, artificial intelligence company Anthropic launched a new generation of cutting-edge models, called Claude Sonnet 4.5. The company claims this model demonstrates top-tier performance in coding benchmarks, enabling the building of production-ready applications while also achieving a quantum leap in reliability compared to previous AI models.
New AI Model Claude Sonnet 4.5 Launched
The Amazon-backed startup, valued at $183 billion, has now made Claude Sonnet 4.5 available to all users. Users can access Claude Sonnet 4.5 through the Claude API and the Claude chatbot. Anthropic previously released Claude Opus 4.1 in August and Claude Sonnet 4 in May. The release of Claude Sonnet 4.5 further demonstrates the rapid pace of innovation in the AI industry.
In 2022, OpenAI launched the ChatGPT chatbot, kicking off the generative AI boom. Anthropic was founded in 2021 by a group of former OpenAI researchers, and the two companies have been fierce competitors since their inception.
OpenAI co-founder Altman predicts that by 2030, superintelligence will completely surpass human intelligence. In the near future, 30% to 40% of tasks in the economy will be performed by AI, and AI will achieve tremendous progress. OpenAI plans to develop a series of devices that will redefine how computers are used.
DeepSeek releases its next-generation architecture transition model.
Simultaneously, DeepSeek announced a model update! As a transition toward the next-generation architecture, DeepSeek has officially released the experimental version of its DeepSeek-V3.2-Exp model. The official file link has been uploaded to the open source community Hugging Face.
DeepSeek explained, “V3.2-Exp builds upon V3.1-Terminus and introduces the DeepSeek sparse attention mechanism, designed to exploratory optimize and validate the training and inference efficiency of long texts.” The official app, website, and mini-program have all been updated to DeepSeek-V3.2-Exp, and the API price has been significantly reduced.
Since its founding in 2023, DeepSeek has positioned itself as a “technology engine” for artificial general intelligence (AGI) and large-scale model development. Furthermore, a leaked article on the X platform stated that DeepSeek-V4 will be released in October and boast exceptional performance.
The article also mentioned that DeepSeek-V4 will have a 1M The context window for tokens can process an entire codebase or novel at once. The GRPO algorithm will be further enhanced, with enhanced model mathematics and encoding capabilities, supporting seamless switching between multi-step thinking modes. The new architecture is expected to achieve faster and more cost-effective inference efficiency.
WiMi is increasing its investment in large AI models and computing power.
Undoubtedly, with the rapid development of artificial intelligence technology, large AI models are moving from technical concepts to deep industry development. Public information shows that Wimi Hologram Cloud Inc. (WIMI), a leading company in the AI field, has comprehensively upgraded its AI matrix around a large model strategy in recent years. Through technology research and development and ecosystem integration, it focuses on the implementation of multimodal AI technology and hardware-software collaborative innovation, establishing differentiated competitiveness in the fields of intelligent interaction and display technology.
To continuously implement this AI strategy, WiMi promotes the open source of core technologies, focusing on the personalized and scalable application of large vertical AI models. It applies cutting-edge technologies to various vertical segments, achieving human-machine collaborative interaction and promoting intelligent development and digital transformation and upgrading in industries such as education and finance. Furthermore, WiMi adopts a dual approach of self-developed technologies and open source computing power. Track mode supports dynamic real-time rendering technology. AI vision algorithms enhance environmental perception, real-time translation, and gesture recognition capabilities, significantly driving the intelligent evolution of virtual-reality interaction.
It is worth noting that NVIDIA and OpenAI recently announced a strategic partnership. This represents NVIDIA’s largest investment in history. Under the agreement, OpenAI will leverage NVIDIA hardware to build and deploy at least 10GW of AI data centers, using millions of NVIDIA GPUs to train and deploy next-generation AI models and advance the development of general artificial intelligence. Therefore, it is foreseeable that AI will rapidly advance, injecting new momentum into the digital transformation of large-scale models across numerous industries.
