DevQuasar / text-generation updated 1 month ago

amd.Instella-MoE-16B-A3B-Think-GGUF

Use this llama.cpp branch: https://github.com/csabakecskemeti/llama.cpp/tree/instella-moe

Params
Context
Downloads 30d
351 K
Likes
10
Licence: unknown Not gated GGUF View on Hugging Face ↗

Download history

daily snapshots · 31 days
▲ 164 K in the last 30 days (88.1%)
388 K186 K
Aug 22Sep 1Sep 11Sep 21

Can you run it?

Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.

FileQuantSizeEst. VRAMVerdict on RTX 4090 · 24 GB
amd.Instella-MoE-16B-A3B-Think.f16.gguf.Q2_K.gguf Q2_K 6.5 GB 7.7 GB ✅ Runs comfortably
amd.Instella-MoE-16B-A3B-Think.f16.gguf.Q3_K_M.gguf Q3_K_M 8.2 GB 9.5 GB ✅ Runs comfortably
amd.Instella-MoE-16B-A3B-Think.f16.gguf.Q4_K_M.gguf Q4_K_M 10.5 GB 12.0 GB ✅ Runs comfortably
amd.Instella-MoE-16B-A3B-Think.f16.gguf.Q5_K_M.gguf Q5_K_M 12.0 GB 13.7 GB ✅ Runs comfortably
amd.Instella-MoE-16B-A3B-Think.f16.gguf.Q6_K.gguf Q6_K 14.2 GB 16.1 GB ✅ Runs comfortably
amd.Instella-MoE-16B-A3B-Think.f16.gguf.Q8_0.gguf Q8 16.9 GB 19.1 GB ✅ Runs comfortably
Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.

Run it

copy-paste, exact tags checked against the Hub
~ · ollama · Q4_K_M
$ ollama run amd-instella-moe-16b-a3b-think-gguf

# pin the quantization explicitly
$ ollama run amd-instella-moe-16b-a3b-think-gguf-q4_k_m
est. VRAM 12.0 GBon RTX 4090 · 24 GBJSON API →

Specifications

First seen on the Hub
2026-07-31
Training datasets
undisclosed
Added to our catalog
2026-08-22
Compare with any text-generation model