family member

Phi 4 MINI

21 builds of Phi 4 MINI measured on real machines — different packagers, precisions and engines, each a different file with its own numbers. The best each engine proved is first; every build has its own page with the machines and signatures behind it.

on ollama · 20 builds

trained 128K · proved 128K context proved by maziyarpanahi Q3_K_L

154.3 tok/s peak, by maziyarpanahi Q2_K

on vllm · 1 build

proved 4K context proved by microsoft BF16

proved by at least one build: Structured output · Sustained output · Writes code · Reasons about code · Synthesises sources · Follows instructions

buildpublisherprecisionenginewhose workcontext provedbest tok/smachines
hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_LmaziyarpanahiQ3_K_Lollama—1,31,072110.13
hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q4_K_MunslothQ4_K_Mollamaquantizer1,31,07275.32
hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q5_K_MunslothQ5_K_Mollamaquantizer1,31,07268.71
hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q3_K_MunslothQ3_K_Mollamaquantizer1,31,07266.91
hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q8_0unslothQ8_0ollamaquantizer1,31,07253.71
hf.co/unsloth/Phi-4-mini-instruct-GGUF:BF16unslothBF16ollamaquantizer1,31,07232.02
hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q3_K_LbartowskiQ3_K_Lollamaquantizer8,19232.11
hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_SmaziyarpanahiQ3_K_Sollama—4,096134.83
hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q4_K_MbartowskiQ4_K_Mollamaquantizer4,09640.61
hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ2_MbartowskiIQ2_Mollamaquantizer4,09637.32
hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ4_XSbartowskiIQ4_XSollamaquantizer4,09636.21
hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q4_0bartowskiQ4_0ollamaquantizer4,09633.71
hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ3_XXSbartowskiIQ3_XXSollamaquantizer4,0960.71
microsoft/Phi-4-mini-instructmicrosoftBF16vllmvendor4,096—1
hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q2_KmaziyarpanahiQ2_Kollama——154.31
hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q4_K_MmaziyarpanahiQ4_K_Mollama——101.91
hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q4_K_SmaziyarpanahiQ4_K_Sollama——77.41
hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q2_KunslothQ2_Kollamaquantizer—71.01
hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q8_0maziyarpanahiQ8_0ollama——39.21
hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q3_K_SbartowskiQ3_K_Sollamaquantizer—28.21
hf.co/Mungert/Phi-4-mini-instruct.gguf:IQ2_XSmungertIQ2_XSollama——8.51
derivatives — somebody else's change to these weights

Fine-tunes, merges and abliterations of this member, each its own model with its own page. None of their numbers are counted above.