DiggTop ⌕

Carnice-9b

Qwen family · 9.0B params

1recorded run
56.1best tok/s
1verified

Evidence: 1 sourced run · 1 independent source · first recorded Jul 5, 2026. What these labels mean.

What the Carnice-9b numbers mean

The fastest verified result on our board is 56.1 output tok/s on an RTX 3060 x2 at Q4_K_M with a 512-token context. That is about 17.8 ms per generated token — 500 tokens takes roughly 8.9 seconds. Prompt processing on the same record runs at 1621.8 tok/s, so a 2,000-token prompt is read in about 1.2 seconds. Peak VRAM was measured at 6.7 GB.

How much VRAM does Carnice-9b need?

Carnice-9b has 9.0B parameters. These estimates use the same formula as the VRAM calculator: weights, plus about 15% for a moderate-context KV cache and runtime overhead, plus a small engine floor.

QuantisationEstimated VRAMSmallest card that fits
Q4_K_M (4-bit)6.6 GB8 GB
Q8_0 (8-bit)11.8 GB16 GB
FP16 (16-bit)21.5 GB32 GB

Long contexts and large batches raise the real number — read the smallest card as "possible with a reduced context". Measured throughput is on the benchmark board.

Fastest GPUs for this model (best tok/s)
  1. RTX 3060 x256.1

Models with a similar footprint

ModelParamsQ4_K_M weights
DeepSeek-V4-Pro-Qwen3.5-9B-MTP9B5.1 GB
Huihui-Qwythos-9B-Claude-Mythos-5-1M-abliterated-i19B5.1 GB
NVIDIA-Nemotron-Nano-9B-v29B5.1 GB
Ornith-1.0-9B9B5.1 GB
Qwen3-VL-8B-Instruct9B5.1 GB
Qwen3.5-9B-MTP9B5.1 GB

Ranked by distance from Carnice-9b's parameter count. Full benchmark board.

All benchmarks Methodology

GPUModelQuantBackendCtx tok/sprompt tok/sVRAM GBSource
RTX 3060 x2 · 12 GBCarnice-9bQ4_K_Mllama.cpp 2d9736351256.1 ✓1621.86.7src
↑