AI

Microsoft confirms OpenAI has been using Looped Transformers in the GPT-6 series

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 06/10/2026, 15:47:26 UTC.

Briefings

Microsoft publicly confirmed that OpenAI utilizes Looped Transformers in its GPT-6 series, validating The Information's prior reporting. The GPT-6.1 model employs exactly two inference passes, a reduction from the previously reported three passes. Microsoft clarified that GPT-6.1 is a post-trained variant of the same base model as GPT-6 Sol, differing only by one fewer loop iteration rather than identical final weights.

Key points

• GPT-6.1 utilizes exactly two inference passes instead of the previously reported three passes.
• GPT-6.1 is a post-trained variant of the same base model as GPT-6 Sol, differing by one fewer loop iteration.

Context

Microsoft publicly confirmed OpenAI's use of Looped Transformers in the GPT-6 series, validating The Information's reporting. This clarification distinguishes between GPT-6.1 and GPT-6 Sol, which share the same base model but differ in post-training loops.

Limitations

Evidence is limited to Microsoft's public web page update; no internal model weights or training data are provided. The specific number of inference passes for other GPT-6 variants remains unconfirmed.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 21/10/2026, 23:21:52 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeic3y2vitqpk4l5dcckgz4rmhngvthjbpruuco6mlfglgrgcbhnh5a