OpenAIOpenAI　·　当前代current generation gpt-ossgpt-oss（2025-08）

gpt-ossgpt-oss模型家族谱系卡model family lineage card

OpenAI 首次放出的开放权重 MoE：Apache-2.0、官方就按「量化后交付」设计（MXFP4），单卡可跑是它的主打场景。本站收录 gpt-oss 系列的量化对比。OpenAI's first open-weight MoE line: Apache-2.0, designed to ship quantized (MXFP4) from the start, single-GPU operation is its headline scenario. This site covers the gpt-oss quantisation comparison.

本站已收录该家族的量化对比：This site covers this family's quantisations: gpt-ossgpt-oss

先说清楚：这个站的数据怎么来的First, where this site's data comes from

✅ 实测字段Measured fields

直接从 Hugging Face 公开 API 读取的客观数据：文件体积（字节）、上下文长度、架构、许可、下载量、更新时间。Objective data read straight from the Hugging Face public API: file sizes in bytes, context length, architecture, licence, downloads, last updated.

✅ 推导Derived

由实测字段按公开规则算出：每权重比特数（依 llama.cpp 命名标准）、档位归属、显存需求区间。推导规则在Computed from the measured fields by published rules: bits per weight (per the llama.cpp naming convention), tier assignment, VRAM requirement range. The rules are fully published in METHODOLOGY.md 完整公开，任何人可复算。 — anyone can recompute them.

❌ 我们不说的What we do not claim

哪个量化质量更高、哪个更快、哪个模型更强——这些需要实测，本站不跑 benchmark，所以不装作知道。凡我们给出的主观判断，一律显式标注「我们的口径」。Which quantisation is higher quality, which is faster, which model is stronger — these need measurement. This site does not run benchmarks, so it does not pretend to know. Every subjective judgement we do offer is explicitly marked as our own reading.

本站This site does 不下载、不托管任何模型权重文件not download or host any model weight files。所有数字都来自公开元数据，点击即可回到发布页核对。. Every figure comes from public metadata, and clicking it takes you back to the release page to verify. 采集时间Collected: 2026-10-30（数据快照，非实时） (a snapshot, not live)

中文生态China availability 国产 / 国内可直接用 / 备案状态 · 快照domestic / direct access / CAC filing · snapshot 2026-10-20

出品属地Origin　国外出品foreign-made　出品方为 OpenAI。Made by OpenAI.
国内下载Direct access　直连不可达direct blocked　Hugging Face 直连不可达；ModelScope 检索 gpt-oss 未命中镜像（2026-10-20）。Hugging Face direct unreachable; no gpt-oss mirror found on ModelScope (2026-10-20).
备案状态CAC filing　未见于备案名单not in filing list：公示名单未检索到 OpenAI / GPT 相关条目（2026-10-20）。No OpenAI / GPT entry found in the public registry (searched 2026-10-20).

备案信息以网信办《生成式人工智能服务已备案信息》公示名单为准（按批次滚动更新，本站只抄编号不解读）；「国内可直连」是本站从大陆网络直连实测的结果，不是厂商承诺。Filing status is quoted from the CAC public registry (updated batch by batch; this site copies the number and does not interpret it). "Direct access" is this site's own reachability measurement from a mainland network, not a vendor promise.

网信办《生成式人工智能服务已备案信息》滚动公告页（13 个批次附件 PDF，2026-10-20 全量下载检索）：CAC · filed generative-AI service information (rolling announcement page, 13 batch attachment PDFs, full download & search on 2026-10-20): https://www.cac.gov.cn/2024-04/02/c_1713729983803145.htm
ModelScope 站内检索 gpt-oss（2026-10-20，无结果）：ModelScope on-site search for gpt-oss (2026-10-20, no results): https://www.modelscope.cn/search?search=gpt-oss

代次Generations

gpt-ossgpt-oss2025-08当前代current

唯一一代（2025-08 发布）。官方 README：120B 版 117B 总参 / 5.1B 激活，20B 版 21B 总参 / 3.6B 激活；MoE 权重以 MXFP4 后训练量化，120B 单张 80GB 卡可跑、20B 16GB 内可跑。Sole generation (released 2025-08). Official README: the 120B build has 117B total / 5.1B active parameters, the 20B build 21B total / 3.6B active; MoE weights post-trained with MXFP4 quantization — 120B fits one 80GB GPU, 20B fits within 16GB.
官方页面 ↗official page ↗

尺寸档Size tiers

本地 20B（21B 总参 / 3.6B 激活，16GB 内）Local 20B (21B total / 3.6B active, within 16GB)
单卡服务器 120B（117B 总参 / 5.1B 激活，80GB 卡）Single-GPU server 120B (117B total / 5.1B active, 80GB card)

许可Licences

apache-2.0Apache-2.0（官方 README：「Permissive Apache 2.0 license」，可自由商用）Apache-2.0 (official README: "Permissive Apache 2.0 license", free for commercial use) 许可原文 ↗licence text ↗

官方能力声明（带出处）Official capability claims (with sources)

官方 README：两档规格与激活参数（117B/5.1B、21B/3.6B）、MXFP4 量化后训练、可配置推理强度（low/medium/high）、支持微调。Official README: the two size builds with active-parameter counts (117B/5.1B, 21B/3.6B), MXFP4 post-training quantization, configurable reasoning effort (low/medium/high), fine-tunable.出处 ↗source ↗

以上均为厂商官方页面原文或官方自评基准的转述；涉及横评结论的，本站不背书——本站不跑 benchmark。The above are quotations of official pages or transcriptions of vendor-run benchmarks; this site does not endorse cross-model conclusions — it runs no benchmarks.

出处Sources

gpt-oss-120b 模型页与 README（2026-10-20 直抓 200）：gpt-oss-120b model page & README (direct fetch 200 on 2026-10-20): https://huggingface.co/openai/gpt-oss-120b
gpt-oss-20b 模型页（HF API 2026-10-20：apache-2.0）：gpt-oss-20b model page (HF API 2026-10-20: apache-2.0): https://huggingface.co/openai/gpt-oss-20b

核验快照：2026-10-30 · 本站不实测、不排名，未知就写未知。Verification snapshot: 2026-10-30 · no benchmarks, no rankings — unknown stays unknown.