SMRTR AIAug 5, 2025Interesting Engineering

New gpt-oss model from NVIDIA and OpenAI hits record 1.5M tokens per second

SMRTR summary

OpenAI and NVIDIA have released two powerful open-weight language models, gpt-oss-120b and gpt-oss-20b, that achieve record speeds of 1.5 million tokens per second on NVIDIA's hardware. These Apache 2.0-licensed models deliver advanced reasoning capabilities comparable to proprietary systems, with the larger version matching OpenAI's o4-mini while the smaller model runs efficiently on devices with just 16GB of memory, making cutting-edge AI accessible to developers worldwide.

SMRTR provides this summary for quick context. The original article belongs to Interesting Engineering.

Read the original article
SMRTR AI

Get the next batch of curated summaries in your inbox.

This archive is built from SMRTR newsletter summaries. Subscribe for hand-picked stories without the extra noise.