OpenAI Debuts First AI Model Optimized for Cerebras Chips, Reducing Nvidia Reliance
The ChatGPT maker balances Nvidia loyalty with a wider supplier strategy.
News
- AI Companies Gear To Tackle Extremism With An Intervention Tool
- Iran Deploys ChatGPT in Growing Cyber Offensive Against UAE, Says Official
- IPO Move by SpaceX Sets Stage for Musk to Become World’s 1st Trillionaire
- Iran Attacks Amazon’s Cloud Business in Bahrain, Report Claims
- OpenAI Raises $122B in Funding Round, Valuation Hits $852B
- IRGC Threatens 18 US Firms as Middle East Conflict Deepens
OpenAI has introduced its first AI model optimised for Cerebras Systems Inc. chips, signalling a significant shift in its hardware approach as it moves away from its longtime dependence on Nvidia Corp.
The model, GPT-5.3-Codex-Spark, is a streamlined, faster variant of OpenAI’s latest Codex system for automating software development tasks. Branded as “Spark,” the model is built for responsiveness rather than raw power. It is optimized for short, iterative coding workflows, such as editing specific code segments, running tests, and quickly pivoting between tasks.
Engineers can interrupt processes midstream and redirect the model without waiting for lengthy computations to conclude, a design choice that prioritizes developer productivity over maximum model scale.
The release follows OpenAI’s recently signed agreement (reportedly valued at more than $10 billion) to use Cerebras hardware to accelerate inference workloads. Cerebras, known for its wafer-scale engine architecture, has positioned itself as an alternative to Nvidia’s graphics processing units in AI training and inference. For the startup, the partnership represents a strategic validation in a semiconductor market where Nvidia has set the pace in both performance and ecosystem integration.
For OpenAI, the move reflects a broader effort to diversify its compute stack amid surging demand for AI services. The company has pursued multiple supplier relationships in recent months. In October, it struck a large-scale deal with Advanced Micro Devices Inc. to deploy 6 gigawatts’ worth of AMD GPUs over several years. It also agreed to purchase custom chips and networking components from Broadcom Inc.
At the same time, OpenAI has sought to reaffirm its ties with Nvidia following reports of friction between the two firms. In a statement accompanying the Spark launch, an OpenAI spokesperson described Nvidia as “foundational” to its infrastructure, noting that its most powerful AI models are the product of multi-year hardware–software co-engineering between the companies. Nvidia, the spokesperson added, remains the anchor of OpenAI’s training and inference stack, even as the company expands its supplier ecosystem.
Strategically, the Spark release also targets intensifying competition in AI-assisted coding. Rivals, including Alphabet Inc.’s Google and Anthropic PBC, are vying for dominance in developer tools. OpenAI says Codex now has more than one million weekly active users.
GPT-5.3-Codex-Spark will debut as a research preview for ChatGPT Pro subscribers, with a broader rollout planned in the coming weeks.
