Gemma 3 1B Instruct quants compared
26 GGUF builds by real file size, probed from unsloth/gemma-3-1b-it-GGUF on Hugging Face (2026-09-02). 1B params.
Download Gemma 3 1B Instruct Q4_K_M (0.75 GB) — it fits 8 GB of memory with 16k context. With Ollama: ollama run gemma3:1b
Every Gemma 3 1B Instruct quant by real file size
| Quant | Weights | + KV (16k) | Total | Fits comfortably in | Quality |
|---|---|---|---|---|---|
| BF16 | 1.87 GB | 1.8 GB | 3.6 GB | 8 GB | Full precision (lossless) |
| Q8_K_XL | 1.38 GB | 1.8 GB | 3.1 GB | 8 GB | Near-lossless |
| Q8_0 | 1 GB | 1.8 GB | 2.8 GB | 8 GB | Near-lossless |
| Q6_K_XL | 0.96 GB | 1.8 GB | 2.7 GB | 8 GB | Excellent |
| Q6_K | 0.94 GB | 1.8 GB | 2.7 GB | 8 GB | Excellent |
| Q5_K_XL | 0.81 GB | 1.8 GB | 2.6 GB | 8 GB | Very high |
| Q5_K_M | 0.79 GB | 1.8 GB | 2.5 GB | 8 GB | Very high |
| Q5_K_S | 0.78 GB | 1.8 GB | 2.5 GB | 8 GB | Very high |
| Q4_K_M * | 0.75 GB | 1.8 GB | 2.5 GB | 8 GB | High — the default pick |
| Q4_K_XL | 0.75 GB | 1.8 GB | 2.5 GB | 8 GB | High |
| Q4_K_S | 0.73 GB | 1.8 GB | 2.5 GB | 8 GB | High |
| Q4_1 | 0.71 GB | 1.8 GB | 2.5 GB | 8 GB | High |
| IQ4_NL | 0.67 GB | 1.8 GB | 2.4 GB | 8 GB | High |
| IQ4_XS | 0.67 GB | 1.8 GB | 2.4 GB | 8 GB | High |
| Q4_0 | 0.67 GB | 1.8 GB | 2.4 GB | 8 GB | High |
| Q3_K_XL | 0.68 GB | 1.8 GB | 2.4 GB | 8 GB | Acceptable — visible loss |
| Q3_K_M | 0.67 GB | 1.8 GB | 2.4 GB | 8 GB | Acceptable — visible loss |
| Q3_K_S | 0.64 GB | 1.8 GB | 2.4 GB | 8 GB | Acceptable — visible loss |
| IQ3_XXS | 0.55 GB | 1.8 GB | 2.3 GB | 8 GB | Acceptable — visible loss |
| Q2_K_XL | 0.65 GB | 1.8 GB | 2.4 GB | 8 GB | Experimental — not ranked — never recommended |
| Q2_K | 0.64 GB | 1.8 GB | 2.4 GB | 8 GB | Experimental — not ranked — never recommended |
| Q2_K_L | 0.64 GB | 1.8 GB | 2.4 GB | 8 GB | Experimental — not ranked — never recommended |
| IQ2_M | 0.54 GB | 1.8 GB | 2.3 GB | 8 GB | Experimental — not ranked — never recommended |
| IQ2_XXS | 0.53 GB | 1.8 GB | 2.3 GB | 8 GB | Experimental — not ranked — never recommended |
| IQ1_M | 0.52 GB | 1.8 GB | 2.3 GB | 8 GB | Experimental — not ranked — never recommended |
| IQ1_S | 0.52 GB | 1.8 GB | 2.3 GB | 8 GB | Experimental — not ranked — never recommended |
* default pick. Weights = real GGUF file sizes from unsloth/gemma-3-1b-it-GGUF (probed 2026-09-02). KV = fp16 estimate; a q8_0 cache roughly halves it. "Comfortable" = weights + KV within 90% of memory.
Best Gemma 3 1B Instruct quant by memory
| Memory | Recommended quant | Total (16k ctx) |
|---|---|---|
| 8 GB | BF16 | 3.6 GB |
Why we don't rank Gemma 3 1B Instruct's 2-bit quants
Quants at 2 bits per weight or below (Q2_K, IQ2, IQ1, TQ1) cut file size by roughly half versus Q4, but the quality collapse is steep and non-linear: perplexity spikes, instruction-following degrades, and hallucinations rise. A model that answers faster but wrong is not a smaller model — it is a worse one. ModelFit lists these builds for completeness but never ranks or recommends them.
Frequently asked questions
What is the best quantization of Gemma 3 1B Instruct?
Q4_K_M is the default pick: 0.75 GB of weights, high — the default pick quality, fitting comfortably in 8 GB of memory (weights + 16k context KV-cache). Go Q6_K or Q8_0 if you have headroom.
How much memory does Gemma 3 1B Instruct need?
At Q4_K_M, Gemma 3 1B Instruct needs 0.75 GB for the weights plus ~1.8 GB of KV-cache at 16k context — about 2.5 GB total, so a 8 GB card or Mac (90% usable budget) runs it comfortably.
Should I use a Q2_K or IQ2 quant of Gemma 3 1B Instruct?
No. Gemma 3 1B Instruct at 2 bits per weight is a visibly worse model — quality collapse at that bitrate is steep, not gradual. If only a 2-bit build fits your memory, run a smaller model at Q4_K_M instead. ModelFit lists these builds but never recommends them.
Cite this page
ModelFit: Gemma 3 1B Instruct quantization comparison (real GGUF file sizes). https://modelfit.io/quant-compare/gemma-3-1b/ (data probed 2026-09-02, CC BY 4.0).