OpenAI: gpt-oss open-weight reasoning and tool-use resource

What changed

Released open-weight expert reasoning models with quantized expert weights, variable-effort reasoning, tool-use training and accompanying inference implementations and tokenizers.

Scientific scope

Released open-weight expert reasoning models with quantized expert weights, variable-effort reasoning, tool-use training and accompanying inference implementations and tokenizers. One original two-size system/resource contribution, not new invention of MoE, GQA, RoPE or quantization. Weight memory figures are not complete runtime resource requirements; visible reasoning and default evaluations do not guarantee safety or faithful internal explanations.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. One original two-size system/resource contribution, not new invention of MoE, GQA, RoPE or quantization. Weight memory figures are not complete runtime resource requirements; visible reasoning and default evaluations do not guarantee safety or faithful internal explanations.

Contributors

• Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.

• gpt-oss-120b & gpt-oss-20b Model Card
• OpenAI: gpt-oss open-weight reasoning and tool-use resource — official research record

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction