RTX 4080

12 records across 3 models (quants: MXFP4, Q4_K).

hardwaremodelquantbackendtok/spp tok/sWttft sctx/tgdateclassflagssource
RTX 4080gpt-oss 20B 20BMXFP4llama.cpp136.56218.5––4096–sourced–source
RTX 4080gpt-oss 20B 20BMXFP4llama.cpp120.24329.2––16384–sourced–source
RTX 4080gpt-oss 20B 20BMXFP4llama.cpp106.02831.1––32768–sourced–source
RTX 4080gpt-oss 20B 20BMXFP4llama.cpp81.31559.0––65536–sourced–source
RTX 4080gpt-oss 20B 20BMXFP4llama.cpp60.0898.5––131072–sourced–source
RTX 4080Qwen3 14B 14BQ4_Kllama.cpp62.03574.7––4096–sourced–source
RTX 4080Qwen3 14B 14BQ4_Kllama.cpp51.22295.1––16384–sourced–source
RTX 4080Qwen3 14B 14BQ4_Kllama.cpp40.61395.7––32768–sourced–source
RTX 4080Qwen3 8B 8BQ4_Kllama.cpp102.76177.9––4096–sourced–source
RTX 4080Qwen3 8B 8BQ4_Kllama.cpp77.93809.8––16384–sourced–source
RTX 4080Qwen3 8B 8BQ4_Kllama.cpp59.01968.4––32768–sourced–source
RTX 4080Qwen3 8B 8BQ4_Kllama.cpp39.0937.9––65536–sourced–source

All rows cite their source. Provenance: community 40018 sourced 1085

← all hardware