Google DeepMindGoogle DeepMind　·　当前代current generation Gemma 4Gemma 4（2026-03）

GemmaGemma模型家族谱系卡model family lineage card

尺寸档最全的家族（E2B 到 31B），端侧到服务器一条线；许可是 Apache-2.0 + 使用政策组合，下载前读政策原文。本站收录的 gemma-4 系列覆盖其中三档。The family with the widest size spread (E2B to 31B), one line from edge to server. Licensing is Apache-2.0 plus a use policy — read the policy text before download. This site's gemma-4 series covers three of the tiers.

本站已收录该家族的量化对比：This site covers this family's quantisations: Gemma 4Gemma 4

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 Google。Made by Google.
国内下载Direct access　国内可直连direct OK　ModelScope 有 google 官方组织镜像（2026-10-20 实测 200）；Hugging Face 直连不可达。Official google org mirror on ModelScope (measured 200); Hugging Face direct unreachable.　·　国内镜像入口domestic mirror
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 Google / Gemma 相关条目（2026-10-20）。No Google / Gemma 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 官方组织镜像（2026-10-20 实测 200）：ModelScope official-org mirror (200 verified on 2026-10-20): https://www.modelscope.cn/models/google/gemma-4-31B-it

代次Generations

Gemma 4Gemma 42026-03当前代current

当前代。官方口径五档尺寸：E2B / E4B / 12B / 26B-A4B / 31B；dense 与 MoE 两种架构。Current generation. Officially five sizes: E2B / E4B / 12B / 26B-A4B / 31B, in both dense and MoE architectures.
官方页面 ↗official page ↗
Gemma 3Gemma 32025

上一代，1B–27B 多模态小杯路线；Gemma 4 的许可证与使用政策框架在这一代确立。Previous generation, multimodal small builds from 1B to 27B; the licence-plus-use-policy framework carried into Gemma 4 was established here.
官方页面 ↗official page ↗

尺寸档Size tiers

端侧 E2B / E4B：手机与浏览器可跑On-device E2B / E4B: phones and browsers
笔电 12B / 26B-A4BLaptop 12B / 26B-A4B
服务器 31B（密集）Server 31B (dense)

许可Licences

apache-2.0Apache-2.0（附 Google 使用政策，链接以官方许可页为准）Apache-2.0 (subject to Google's use policy; see the official licence page) 许可原文 ↗licence text ↗

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

官方 README：「Gemma 4 features a context window of up to 256K tokens and maintains multilingual support in over 140 languages」——中模型 256K、小模型 128K（官方分区口径）。Official README: "Gemma 4 features a context window of up to 256K tokens and maintains multilingual support in over 140 languages" — 256K on medium builds, 128K on small ones per the official breakdown.出处 ↗source ↗
官方 README：多模态覆盖 text + image（全部规格），video 与 audio 在 E2B / E4B / 12B 上原生支持。Official README: multimodal coverage of text + image on all builds, with video and audio native on E2B / E4B / 12B.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

gemma-4-31B-it 模型页与 README（2026-10-20 直抓 200）：gemma-4-31B-it model page & README (direct fetch 200 on 2026-10-20): https://huggingface.co/google/gemma-4-31B-it
Gemma 4 许可页（官方 license_link 字段指向，2026-10-20 直抓 200）：Gemma 4 licence page (pointed to by the official license_link field, direct fetch 200 on 2026-10-20): https://ai.google.dev/gemma/docs/gemma_4_license
gemma-3-27b-it 模型页（代次核验，2026-10-20 直抓 200）：gemma-3-27b-it model page (generation check, direct fetch 200 on 2026-10-20): https://huggingface.co/google/gemma-3-27b-it

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.