COLLECTION / RECIPES

Recipes

One artifact × hardware × engine unit. Validated rows are launch-safe; measured speed sweeps attach to each recipe as evidence.

20 records

GeForce RTX 50701 × 12 GBGemma-4-12B-itQ4_K_Mllama.cppreference2Kcontext72.1 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBGemma-4-12B-itUD-Q5_K_XLllama.cppreference2Kcontext62.8 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBGemma-4-12B-itUD-Q6_K_XLllama.cppreference2Kcontext51.3 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBGemmable-4-12B-MTP-GGUFQ4_K_Mllama.cppreference2Kcontext71.2 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBGrug-12BQ4_K_Mllama.cppreference2Kcontext69.2 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBMellum2-12B-A2.5B-InstructQ4_K_Mllama.cppreference2Kcontext261.4 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBMellum2-12B-A2.5B-ThinkingQ4_K_Mllama.cppreference2Kcontext265.5 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBNVIDIA-Nemotron-3-Nano-4B-FP8Q4_K_Mllama.cppreference32Kcontext185.8 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBOrnith-1.5-9BQ4_K_Mllama.cppreference2Kcontext102 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBQwen3-14BQ4_K_Mllama.cppreference2Kcontext66.6 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBQwen3.5-35B-A3BIQ1_Mllama.cppreference2Kcontext182.3 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBQwen3.6-35B-A3BQ4_K_Mllama.cppreference62.5Kcontext64.5 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBQwen3.8-27BAD-IQ2-XSllama.cppreference2Kcontext46.1 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBQwen3.8-27BUD-IQ2_XXSllama.cppreference2Kcontext46.3 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 50701 × 12 GBQwen3.8-27BUD-Q2_K_XLllama.cppreference8Kcontext70.8 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 5070 Ti1 × 16 GBLFM2.5-8B-A1B-BaseQ4_K_XLllama.cppreference4Kcontext402.3 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 5070 Ti1 × 16 GBLFM2.5-8B-A1B-BaseQ8_0llama.cppreference4Kcontext313.4 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 5070 Ti1 × 16 GBLFM2.5-8B-A1B-BaseQ8_K_XLllama.cppreference4Kcontext301.3 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 5070 Ti1 × 16 GBQwen3.5-9BQ4_K_Sllama.cppreference8Kcontext124.1 tok/smeasuredcandidate llama.cpp evidence measured

GeForce RTX 5070 Ti1 × 16 GBQwen3.6-35B-REAP-Pruned-ratio-0.5NVFP4vllmreferenceUnknown contextcontext133.3 tok/smeasuredcandidate vllm evidence measured