{"v":1,"id":"model:hf:yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF","slug":"model-yuxinlu1-gemma-4-12b-agentic-fable5-composer2-5-v2-3-5x-tau2-gguf","kind":"model","category":"llm","title":"gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF","summary":"💻🤖 Gemma4-12B v2 — Coding + Agentic Edition ✨ ### 🐣 Tiny footprint, big brain — a local coding & tool-using agent for everyone","source":{"provider":"hf","ref":"yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF","url":"https://huggingface.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF","rev":"190a31365a6b80a692349be34ccdac730cad4fe4","fetchedAt":"2026-10-02T20:52:34.434Z","etag":"W/\"52e0-bghZN/2OS4p1e/YBaF/Zjy0g1HQ\""},"author":{"name":"yuxinlu1","url":"https://huggingface.co/yuxinlu1"},"license":{"spdx":"apache-2.0","raw":"apache-2.0","open":true},"metrics":{"downloads":710251,"downloadsWeek":327838,"likes":1632,"takenAt":"2026-10-02T20:52:34.434Z"},"tags":["gguf","gemma4","coding","agentic","terminal","tool-use","reasoning","thinking","llama.cpp","local-llm","text-generation","endpoints_compatible","conversational"],"pipeline":"text-generation","links":{"github":"google-deepmind/gemma","npm":"node-llama-cpp"},"updatedAt":"2026-06-19T09:31:48.000Z","collectedAt":"2026-10-02T20:52:34.434Z","review":{"numbers":["710,251 downloads on Hugging Face","1,632 likes","license apache-2.0","6.9 GB for gemma4-v2-Q4_K_M.gguf","327,838 npm downloads a week for node-llama-cpp","latest node-llama-cpp@3.22.1"],"log":null},"trust":"unlabeled","health":{"status":"alive","checkedAt":"2026-10-02T20:52:34.434Z","http":200},"description":"# 💻🤖 Gemma4-12B **v2** — Coding + Agentic Edition ✨\n### 🐣 Tiny footprint, big brain — a local **coding & tool-using agent** for *everyone*\n\n> **No matter your GPU. No matter your RAM.** With **~4.5 GB** of VRAM *or* unified memory free, you can run your own\n> private, offline coding **agent** right now. 🚀 v2 is the big **agentic** upgrade — it reads, reasons, *uses tools*,\n> and works through multi-step technical tasks before it acts. 🧠🛠️ All local, all yours, no API, no cloud.\n\n---\n\n## 📊 The headline — it works as an agent (tau2-bench)\n\nv2 is built for **coding + agentic** work — writing code, running commands, using tools, debugging, multi-step\ntechnical tasks. The clearest signal is **tau2-bench `telecom`**, an agentic tool-use benchmark whose\n*diagnose → fix → verify* loop mirrors real terminal/debugging work:\n\n| tau2-bench **telecom** · 20 tasks · local, same harness, **all Q8_0** | score |\n|---|---|\n| official `gemma-4-12B-it` (base) | **~15%** |\n| 🟢 **Gemma4-12B v2 (this model)** | **~55%** |\n\n→ Roughly **3.5× higher** than the base model on technical-agentic tasks. 🎯 **Want the full story** — *why* telecom,\n*how* the two models fail differently, the honest caveats, and the trade-offs (including general knowledge)?\n**It's all broken down further below. 👇**\n\n---\n\n## 🚀 Announcements\n\n**📌 Hitting a problem? Please check my pinned discussion first.** **~99% of issues are a client/sampler config, not\nthe weights** — and they have a quick fix there. For example: garbled or **repeating `0000…`** output almost always\nmeans **no repetition penalty** (set `rep_pen 1.1`, `temp 1.0`); and leaked `` / `` tokens mean\nyour front-end isn't parsing Gemma 4's **native tool format** (use llama.cpp `--jinja`). If your question isn't covered,\n**don't hesitate to open a discussion** — I read them and reply as fast as I can. 💬\n\n**📦 No Q2_K this release.** I finished a Q2…\n\nSource: https://huggingface.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF","install":{"kind":"model","hfId":"yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF","gated":false,"format":"gguf","files":[{"name":"MTP/gemma-4-12B-it-MTP-BF16.gguf","size":861520128,"quant":"BF16","sha256":"20b0e5caf9152e816a56f92c702528bffc7a7c930f20c33cf6616ac216998037"},{"name":"MTP/gemma-4-12B-it-MTP-F16.gguf","size":861520128,"quant":"F16","sha256":"ee0987dd6c29dae4c183713f5614814808f284c28f11c739d3ba075e73332ccf"},{"name":"MTP/gemma-4-12B-it-MTP-Q8_0.gguf","size":465109248,"quant":"Q8_0","sha256":"145db9094bc0f85f1701e255a2ed216dcc9800fc8bc8631ad00905b456bd451b"},{"name":"gemma4-v2-Q3_K_M.gguf","size":6087086624,"quant":"Q3_K_M","sha256":"db66c1ab1eced5c89de37689addc8d37242d9815d82c4ddc571a9bd4b834691e"},{"name":"gemma4-v2-Q4_K_M.gguf","size":7381381664,"quant":"Q4_K_M","sha256":"0b9506cab36f7f818e34f9c0f5a3d6568d0b37100f3a3e1092e2eec3c4c96791"},{"name":"gemma4-v2-Q6_K.gguf","size":9786020384,"quant":"Q6_K","sha256":"edb98c934c50c8a5a32074b27f9ccf6e22552619e70245c0f83ccfbea9b1fe61"},{"name":"gemma4-v2-Q8_0.gguf","size":12669645344,"quant":"Q8_0","sha256":"2c20a496baf3e9a3ead59d37c7afe228a863662d58155f360d44eb8b2465cb7f"}],"totalBytes":38112283520,"suggestedFile":"gemma4-v2-Q4_K_M.gguf","requirements":{"ramGb":9,"diskBytes":7381381664,"note":"estimate: suggested file size × 1.15 + 0.5 GB; a real measurement comes with lsh models"},"runWith":["llama.cpp","ollama"],"command":"lsh models install hf:yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF"}}