community pool / model answer

North-Mini-Code-1.0

code · CohereLabs/North-Mini-Code-1.0

101.6 tok/s · model median · 17 runs in the model record.

Reference rigQuantizationMedian resultBasisTTFT (ms)Peak VRAMMax contextEngines
RTX 3060 12GB ×24-bit89.5 tok/smeasured2,03319.6 GB32,768llama.cpp
RTX 3090 24GB4-bit151.5 tok/smeasured-- GB131,072llama.cpp
RTX 5090 32GB6-bit258.4 tok/sreported4926.9 GB8,192llama.cpp

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.