OpenAI seeks public feedback ahead of open-weight model release
OpenAI posted a feedback form and announced developer sessions in San Francisco, Europe and Asia-Pacific rather than a launch date, saying key decisions were still open.
- Open weights & ecosystem
- Minor
Sam Altman announced that OpenAI would release an open-weight language model with reasoning capability “in the coming months” — its first open-weight release since GPT-2 in 2019 — and pointed developers to a feedback page asking what features they wanted and what open models they had already used. Altman said the company still had “some decisions to make,” and that it would host developer events to gather input before settling the design, starting with a San Francisco session within weeks and further sessions planned in Europe and Asia-Pacific.
The move was notable mainly for what it signalled about strategy rather than for any technical content: after two years in which OpenAI had disclosed progressively less about its models’ architecture, training data and weights, the company was reversing course on openness for at least one release, under competitive pressure from open-weight models from Meta, DeepSeek, Mistral and others. Soliciting public input before a release was itself a departure from OpenAI’s usual pattern of announcing finished products.
The model that resulted, gpt-oss, shipped in two sizes under an Apache 2.0 licence that August.