Alibaba's Qwen 3.8 27B model is now available on Cerebras' inference platform, delivering 1500 tokens per second throughput. The deployment marks expanded access to the open-source model on specialized AI hardware.
Cerebras has added Qwen 3.8 27B to its model roster, offering the 27-billion parameter model at 1500 tokens/second inference speed. The Alibaba-developed Qwen model joins other offerings on Cerebras' inference documentation.
Qwen 3.8 27B represents a mid-size option in Alibaba's Qwen family, balancing capability with computational efficiency. Cerebras' deployment leverages the company's specialized wafer-scale processors designed for AI inference workloads.
The 1500 tokens/second rate positions the offering competitively in the inference acceleration market. Availability signals growing adoption of Qwen models across specialized hardware providers beyond cloud infrastructure platforms.
The announcement generated significant discussion on Hacker News, accumulating 210 points and 69 comments, reflecting developer interest in model availability and inference performance metrics.
Los Angeles Unified School District has prohibited students from accessing generative AI tools on school-issued devices. The restriction aims to reduce screen time in classrooms.
OpenAI CEO Sam Altman says the AI industry has poorly communicated the technology's benefits to the public, leaving developers exposed to mounting global backlash.
ChatGPT, Claude, Grok, and Gemini experienced service interruptions at nearly the same time, marking an unusual convergence of downtime across the AI industry's leading platforms.
OpenAI has unveiled GPT-6 Astra, introducing updated deployment safety protocols detailed in a new system card. The release has generated significant discussion among developers on Hacker News.