Yazılar

Orange to Harness OpenAI’s Latest AI Models for African Languages

/in /tarafından

French telecom giant Orange announced plans to leverage OpenAI’s cutting-edge AI models to advance African language technology. Despite the continent’s rich linguistic diversity—over 2,000 languages—the benefits of AI have largely bypassed African languages due to scarce data and limited computing resources, according to researchers from Cornell University and the journal Nature.

Operating in 18 African countries, Orange signed a deal last year with OpenAI to access pre-release AI models and fine-tune large language models for regional African language translation tasks. The company began deploying OpenAI’s Whisper speech model this year for speech recognition but aims to expand into more sophisticated applications with the latest models.

OpenAI’s open-weight models provide publicly accessible parameters, enabling developers like Orange to customize models for specific needs without needing the original training datasets. Orange plans to fine-tune these models using its own collected samples of African languages and roll them out locally.

Steve Jarrett, Orange’s Chief AI Officer, told Reuters the company intends to provide these fine-tuned models free of charge to local governments and public authorities. He emphasized that the initiative serves as a blueprint for bridging the digital divide through AI, fostering collaboration with local startups and communities to elevate African languages as “first-class citizens” in the AI landscape.

Huawei’s AI Lab Denies Copying Alibaba’s Qwen Model Amid Copyright Claims

/in /tarafından

Huawei’s AI research division, Noah Ark Lab, has denied allegations that its Pangu Pro Moe (Mixture of Experts) large language model copied from Alibaba’s Qwen 2.5 14B model. The lab insisted on Saturday that Pangu Pro was independently developed and trained, refuting claims made in a report by an entity named HonestAGI.

HonestAGI published a paper on GitHub claiming “extraordinary correlation” between Huawei’s Pangu Pro Moe and Alibaba’s Qwen model, suggesting that Huawei’s model might have been “upcycled” rather than trained from scratch. The report also raised concerns about potential copyright violations and false claims regarding Huawei’s investment in the model’s training.

In response, Noah Ark Lab stated that their model is not based on incremental training from other manufacturers’ models but instead includes key innovations in architecture and technical features. They highlighted that Pangu Pro is the first large-scale model built entirely on Huawei’s Ascend chips and confirmed adherence to open-source licensing rules for any third-party code used—though they did not specify which open-source models influenced their work.

Alibaba has yet to comment on the allegations, and the identity of HonestAGI remains unknown. The controversy comes amid rising competition in China’s AI sector, which has been accelerated by the release of open-source models like DeepSeek’s R1 and Alibaba’s Qwen family, designed for consumer and chatbot applications. In contrast, Huawei’s Pangu models are primarily applied in government, finance, and manufacturing sectors.

OpenAI Unveils O3 and O4-Mini Models Featuring Advanced Visual Reasoning

/in /tarafından

OpenAI Launches O3 and O4-Mini AI Models With Enhanced Visual Reasoning

OpenAI has unveiled two new AI models—O3 and O4-Mini—designed to push the boundaries of machine reasoning and visual understanding. These models are successors to the earlier O1 and O3-Mini versions and are available to paid ChatGPT users. Highlighted for their visible chain-of-thought (CoT) capabilities, the new models are built to process complex queries involving both text and visual inputs. Their release follows closely on the heels of the GPT-4.1 model series, marking a busy week for the San Francisco-based AI research company.

Announced via a post on X (formerly Twitter), OpenAI described O3 and O4-Mini as their “smartest and most capable” models to date. One of their standout features is enhanced visual reasoning—the ability to interpret and draw inferences from images. This advancement allows the models to extract detailed context, understand spatial relationships, and interpret ambiguous visual data more effectively than their predecessors.

OpenAI also revealed that these are the first models capable of autonomously using all the tools integrated into ChatGPT, such as Python coding, web browsing, file analysis, and image generation. This multi-tool synergy enables the models to handle more dynamic tasks, such as manipulating images (cropping, zooming, flipping), running analytical scripts, or retrieving information even from flawed or low-quality visuals. The potential applications range from reading difficult handwriting to identifying obscure details in images.

In terms of performance, OpenAI claims that both O3 and O4-Mini outperform previous versions—including GPT-4o and O1—on benchmarks like MMMU, MathVista, “VLMs are blind,” and CharXiv. While no comparisons were made with third-party models, these internal benchmarks suggest a notable leap in reasoning and image-based comprehension. As OpenAI continues to iterate, these releases underscore its ongoing focus on building increasingly versatile and intelligent AI systems.