community pool / model answer

DeepSeek-V4-Pro-Qwen3.5-9B-MTP-GGUF

chat · Jackrong/DeepSeek-V4-Pro-Qwen3.5-9B-MTP-GGUF

81.7 tok/s · model median · 2 runs in the model record.

Reference rigQuantizationMedian resultBasisTTFT (ms)Peak VRAMMax contextEngines
RTX 3090 24GB8-bit75.1 tok/sreported836- GB65,536llama.cpp, vllm

Context tested

The table above is the tested context for each exact rig and quantization cell. Empty fields remain empty rather than being filled with an estimate.