Run paperscarecrow/Gemma-4-31B-it-abliterated locally

License: apache-2.0 ⬇ 326,490 ❤ 112
Parameters30.7B
Context262,144

paperscarecrow/Gemma-4-31B-it-abliterated is a very large instruction-tuned chat model with 30.7 billion parameters, built on the gemma4 architecture. It is released under the apache-2.0 license and has been downloaded 326,490 times.

To run paperscarecrow/Gemma-4-31B-it-abliterated locally at a 4,096-token context, its quantized versions need between 20.88 GB (Q4_K_M, lowest quality) and 60.67 GB (F16, highest quality) of memory, weights plus KV cache and a system margin included.

For most users the best balance is Q4_K_M, needing about 20.88 GB. That means paperscarecrow/Gemma-4-31B-it-abliterated fits entirely in the VRAM of a 24 GB GPU or larger, running fully on the GPU.

→ Guide: How much VRAM do you need?

All quantizations

Quant.Bits QualityWeights KVTotal Speed~Verdict
Q4_K_M 4.87 Good 17.4 GB 2.67 GB 20.88 GB 2.9 t/s Offload
Q8_0 8.51 Excellent 30.39 GB 2.67 GB 33.87 GB Insufficient
F16 16.0 Excellent 57.2 GB 2.67 GB 60.67 GB Insufficient

KV cache estimated (architecture unavailable). Speed is a rough estimate bounded by memory bandwidth.

Frequently asked questions

How much VRAM do you need to run paperscarecrow/Gemma-4-31B-it-abliterated?

You need about 20.88 GB of VRAM to run paperscarecrow/Gemma-4-31B-it-abliterated entirely on the GPU using the Q4_K_M quantization (at a 4,096-token context). Smaller quantizations lower the requirement at the cost of quality.

Can I run paperscarecrow/Gemma-4-31B-it-abliterated on an 8 GB GPU?

Partially. paperscarecrow/Gemma-4-31B-it-abliterated only fits on an 8 GB GPU by offloading part of it to system RAM (with Q4_K_M), which runs but is slower.

Can I run paperscarecrow/Gemma-4-31B-it-abliterated on a 16 GB GPU?

Partially. paperscarecrow/Gemma-4-31B-it-abliterated only fits on a 16 GB GPU by offloading part of it to system RAM (with Q8_0), which runs but is slower.

Can I run paperscarecrow/Gemma-4-31B-it-abliterated on a 24 GB GPU?

Yes. With 24 GB of VRAM you can run paperscarecrow/Gemma-4-31B-it-abliterated fully on the GPU using Q4_K_M (about 20.88 GB).

What is the best quantization for paperscarecrow/Gemma-4-31B-it-abliterated?

If memory allows, higher bits-per-weight means better quality. A common sweet spot is a Q4_K_M or Q5_K_M quantization, which keeps most of the quality while roughly halving the memory versus 8-bit. Pick the highest quantization that still fits in your VRAM.