Gemma 4 E4B

203 records across 24 hardware.

hardwaremodelquantbackendtok/spp tok/sWttft sctx/tgdateclassflagssource
M4 MaxGemma 4 E4B oQ4eoMLX238.12473.0––40962026-10-11community–source
M4 MaxGemma 4 E4B oQ4eoMLX232.62443.0––40962026-10-11community–source
M6Gemma 4 E4B oQ4eoMLX199.72385.0––10242026-10-17community–source
M4 MaxGemma 4 E4B oQ4eoMLX183.12421.0––40962026-10-11communitycontradictionsource
M6Gemma 4 E4B oQ4eoMLX167.32386.0––10242026-10-17communitycontradictionsource
M5 MaxGemma 4 E4B oQ4eoMLX150.96817.0––40962026-10-16community–source
M6Gemma 4 E4B oQ4eoMLX136.92697.0––40962026-10-17community–source
M5 MaxGemma 4 E4B oQ4eoMLX135.36589.0––163842026-10-16community–source
M5 MaxGemma 4 E4B oQ4eoMLX131.72918.0––10242026-10-16community–source
M5 MaxGemma 4 E4B oQ4eoMLX124.76992.0––81922026-10-16community–source
M5 ProGemma 4 E4B oQ4eoMLX121.52364.0––10242026-10-16community–source
M4 MaxGemma 4 E4B oQ4eoMLX102.82448.0––40962026-10-11communitycontradictionsource
M4 MaxGemma 4 E4B oQ4eoMLX102.52483.0––40962026-10-11communitycontradictionsource
M4 MaxGemma 4 E4B oQ4eoMLX100.32449.0––40962026-10-11communitycontradictionsource
M4 MaxGemma 4 E4B oQ4eoMLX100.22424.0––40962026-10-11communitycontradictionsource
M4 MaxGemma 4 E4B oQ4eoMLX100.02425.0––40962026-10-11communitycontradictionsource
M6Gemma 4 E4B oQ4eoMLX95.62207.0––10242026-10-12communitycontradictionsource
M5 MaxGemma 4 E4B oQ4eoMLX94.16185.0––327682026-10-16community–source
M4 MaxGemma 4 E4B oQ4eoMLX92.72440.0––40962026-10-11communitycontradictionsource
M2 ProGemma 4 E4B oQ4eoMLX92.7613.4––10242026-10-16community–source
M3 MaxGemma 4 E4B 4bitoMLX92.22087.0––10242026-10-13community–source
Arc B580Gemma 4 E4B Q4_0llama.cpp (Vulkan)90.8––––2026-10-05community–source
Arc B580Gemma 4 E4B Q4_0llama.cpp (Vulkan)90.73323–––2026-10-05community–source
M3 UltraGemma 4 E4B 4bitoMLX89.32798.0––10242026-10-09community–source
Arc B580Gemma 4 E4B Q4_0llama.cpp (Vulkan)87.0––––2026-10-05community–source
Arc B580Gemma 4 E4B Q4_0llama.cpp (Vulkan)86.23449–––2026-10-05community–source
M3 UltraGemma 4 E4B 4bitoMLX85.62949.0––40962026-10-09community–source
M3 UltraGemma 4 E4B 8bitoMLX83.72795.0––10242026-10-09community–source
M3 UltraGemma 4 E4B 8bitoMLX80.62722.0––40962026-10-09community–source
M5 ProGemma 4 E4B nvfp4oMLX78.24118.0––40962026-10-16community–source
M6Gemma 4 E4B oQ4eoMLX76.82444.0––163842026-10-15community–source
M6Gemma 4 E4B oQ4eoMLX72.92683.0––40962026-10-17communitycontradictionsource
M6Gemma 4 E4B oQ4eoMLX72.42626.0––40962026-10-13communitycontradictionsource
M5Gemma 4 E4B oQ4eoMLX72.11690.0––40962026-10-10community–source
M4 ProGemma 4 E4B 4bitoMLX69.31053.0––10242026-10-07community–source
M5Gemma 4 E4B oQ4eoMLX67.91485.0––10242026-10-10community–source
M2 UltraGemma 4 E4B 4bitoMLX67.51943.0––10242026-10-14community–source
M4 ProGemma 4 E4B 4bitoMLX67.41063.0––40962026-10-07community–source
M5 ProGemma 4 E4B nvfp4oMLX66.13981.0––163842026-10-16community–source
M2 UltraGemma 4 E4B 4bitoMLX65.81896.0––40962026-10-14community–source
M4 MaxGemma 4 E4B 4bitoMLX64.71991.0––40962026-10-07community–source
M5Gemma 4 E4B oQ4eoMLX63.41682.0––40962026-10-10communitycontradictionsource
M3 MaxGemma 4 E4B 4bitoMLX61.91919.0––40962026-10-11community–source
M4Gemma 4 E4B oQ4eoMLX60.8649.4––10242026-10-18community–source
M5 ProGemma 4 E4B oQ4eoMLX60.13393.0––40962026-10-16community–source
M3 MaxGemma 4 E4B 4bitoMLX59.11692.0––40962026-10-11community–source
M4 MaxGemma 4 E4B 4bitoMLX58.01932.0––163842026-10-07community–source
M6Gemma 4 E4B oQ4eoMLX58.02589.0––40962026-10-12communitycontradictionsource
M1 MaxGemma 4 E4B 4bitoMLX57.5726.8––10242026-10-08community–source
M4 ProGemma 4 E4B qat-OptiQ-4bitoMLX55.91049.0––10242026-10-07community–source
M5 MaxGemma 4 E4B oQ4eoMLX54.85354.0––655362026-10-16community–source
M4Gemma 4 E4B oQ4eoMLX54.8650.2––10242026-10-16community–source
M1 MaxGemma 4 E4B 4bitoMLX54.2809.7––40962026-10-08community–source
M4 ProGemma 4 E4B qat-OptiQ-4bitoMLX54.11060.0––40962026-10-07community–source
M5 ProGemma 4 E4B nvfp4oMLX53.82507.0––40962026-10-07communitycontradictionsource
M1 MaxGemma 4 E4B 4bitoMLX50.4833.6––81922026-10-08community–source
M2 ProGemma 4 E4B mxfp4oMLX50.0564.6––10242026-10-12community–source
M5 ProGemma 4 E4B 8bitoMLX49.13313.0––40962026-10-16community–source
M5 ProGemma 4 E4B nvfp4oMLX48.53491.0––163842026-10-07communitycontradictionsource
M4 ProGemma 4 E4B 4bitoMLX48.5990.8––10242026-10-07communitycontradictionsource
M2 ProGemma 4 E4B mxfp4oMLX47.8589.9––40962026-10-12community–source
M4 ProGemma 4 E4B 4bitoMLX47.51251.0––10242026-10-17community–source
M4 ProGemma 4 E4B 4bitoMLX47.51251.0––10242026-10-17community–source
M4 ProGemma 4 E4B 4bitoMLX47.31044.0––40962026-10-07communitycontradictionsource
M1 MaxGemma 4 E4B 4bitoMLX47.1680.4––10242026-10-09communitycontradictionsource
M4Gemma 4 E4B oQ4eoMLX46.5641.3––40962026-10-14community–source
M1 MaxGemma 4 E4B 4bitoMLX46.3800.6––40962026-10-17communitycontradictionsource
M1 MaxGemma 4 E4B 4bitoMLX46.3800.6––40962026-10-17communitycontradictionsource
M4Gemma 4 E4B oQ4eoMLX46.2645.5––40962026-10-16community–source
M4Gemma 4 E4B oQ4eoMLX46.0649.3––81922026-10-18community–source
M6Gemma 4 E4B oQ4eoMLX45.91436.0––10242026-10-13communitycontradictionsource
M2 ProGemma 4 E4B oQ4eoMLX45.6629.8––40962026-10-15community–source
M4 ProGemma 4 E4B 4bitoMLX45.41292.0––40962026-10-17community–source
M4 ProGemma 4 E4B 4bitoMLX45.41292.0––40962026-10-17community–source
M1 MaxGemma 4 E4B 4bitoMLX45.2806.0––10242026-10-09communitycontradictionsource
M1 MaxGemma 4 E4B 4bitoMLX45.1809.9––163842026-10-08community–source
M4Gemma 4 E4B oQ4eoMLX44.9655.8––40962026-10-16community–source
M6Gemma 4 E4B oQ4eoMLX44.72327.0––40962026-10-13communitycontradictionsource
M5 ProGemma 4 E4B 8bitoMLX44.73129.0––163842026-10-16community–source
M1 MaxGemma 4 E4B 4bitoMLX44.6784.1––40962026-10-09communitycontradictionsource
M6Gemma 4 E4B oQ4eoMLX44.21613.0––10242026-10-07communitycontradictionsource
M3 MaxGemma 4 E4B 4bitoMLX44.1877.5––10242026-10-16community–source
M1 MaxGemma 4 E4B 4bitoMLX43.6767.9––40962026-10-09communitycontradictionsource
M3 MaxGemma 4 E4B 4bitoMLX43.4931.1––40962026-10-16community–source
M3 ProGemma 4 E4B 4bitoMLX43.2996.1––10242026-10-14community–source
M4 ProGemma 4 E4B 8bitoMLX42.91218.0––10242026-10-16community–source
M4 ProGemma 4 E4B oQ8oMLX42.81252.0––10242026-10-09community–source
M2 ProGemma 4 E4B oQ4eoMLX42.8627.7––40962026-10-16community–source
M6Gemma 4 E4B oQ4eoMLX42.72317.0––40962026-10-07communitycontradictionsource
M5Gemma 4 E4B oQ4eoMLX42.41539.0––10242026-10-10communitycontradictionsource
M4 ProGemma 4 E4B 8bitoMLX42.01287.0––40962026-10-16community–source
M4 ProGemma 4 E4B oQ8oMLX41.91289.0––40962026-10-09community–source
M4 MaxGemma 4 E4B oQ4eoMLX41.62459.0––40962026-10-11communitycontradictionsource
M5Gemma 4 E4B oQ4eoMLX41.61089.0––10242026-10-17communitycontradictionsource
M6Gemma 4 E4B 4bitoMLX40.72502.0––40962026-10-09community–source
M1 ProGemma 4 E4B oQ4eoMLX40.5538.7––10242026-10-08community–source
M6Gemma 4 E4B oQ4eoMLX40.52151.0––327682026-10-15community–source
M5Gemma 4 E4B oQ4eoMLX40.11586.0––10242026-10-10communitycontradictionsource
M5Gemma 4 E4B 4bitoMLX39.62196.0––40962026-10-07community–source
M5Gemma 4 E4B oQ4eoMLX39.61565.0––40962026-10-17communitycontradictionsource
M1 MaxGemma 4 E4B 4bitoMLX39.4811.4––163842026-10-17communitycontradictionsource
M1 MaxGemma 4 E4B 4bitoMLX39.4811.4––163842026-10-17communitycontradictionsource
M5Gemma 4 E4B oQ4eoMLX39.31568.0––10242026-10-10communitycontradictionsource
M5 MaxGemma 4 E4B oQ4eoMLX39.13881.0––1310722026-10-16community–source
M5Gemma 4 E4B oQ4eoMLX38.61661.0––40962026-10-10communitycontradictionsource
M5Gemma 4 E4B oQ4eoMLX38.21633.0––40962026-10-10communitycontradictionsource
M1 ProGemma 4 E4B oQ4eoMLX36.5516.7––40962026-10-08community–source
M3 ProGemma 4 E4B 4bitoMLX35.8774.2––10242026-10-14communitycontradictionsource
M6Gemma 4 E4B oQ4eoMLX35.72138.0––81922026-10-15community–source
M6Gemma 4 E4B 4bitoMLX35.52534.0––163842026-10-09community–source
M4Gemma 4 E4B 4bitoMLX34.9654.4––10242026-10-16community–source
M1 MaxGemma 4 E4B oQ4eoMLX34.0287.9––163842026-10-13community–source
M5Gemma 4 E4B 4bitoMLX33.92045.0––163842026-10-07community–source
M4Gemma 4 E4B 4bitoMLX32.5651.3––40962026-10-16community–source
M4Gemma 4 E4B 4bitoMLX32.2568.8––40962026-10-12community–source
M4Gemma 4 E4B oQ4eoMLX31.9650.7––40962026-10-11communitycontradictionsource
M4Gemma 4 E4B 4bitoMLX31.3652.3––10242026-10-12communitycontradictionsource
M4Gemma 4 E4B 4bitoMLX31.2654.3––10242026-10-14communitycontradictionsource
M4Gemma 4 E4B 4bitoMLX31.2662.6––40962026-10-14community–source
M4Gemma 4 E4B 4bitoMLX31.2654.5––10242026-10-14communitycontradictionsource
M4Gemma 4 E4B 4bitoMLX31.1664.0––40962026-10-12community–source
M4Gemma 4 E4B 4bitoMLX30.9629.1––10242026-10-12communitycontradictionsource
M4Gemma 4 E4B oQ4eoMLX30.7648.8––81922026-10-11communitycontradictionsource
M4Gemma 4 E4B 4bitoMLX30.7662.0––40962026-10-14community–source
M4Gemma 4 E4B oQ4eoMLX30.7602.2––10242026-10-16communitycontradictionsource
M5 ProGemma 4 E4B 8bitoMLX30.62312.0––655362026-10-06community–source
M5 ProGemma 4 E4B oQ4eoMLX30.61652.0––10242026-10-16community–source
M4Gemma 4 E4B 4bitoMLX30.5654.5––40962026-10-12community–source
M2Gemma 4 E4B 4bitoMLX29.1393.9––10242026-10-14community–source
M6Gemma 4 E4B 8bitoMLX28.91908.0––10242026-10-08community–source
M4Gemma 4 E4B 4bitoMLX28.8548.9––163842026-10-12community–source
M6Gemma 4 E4B 8bitoMLX28.22527.0––40962026-10-08community–source
M5Gemma 4 E4B 4bitoMLX27.01800.0––10242026-10-13community–source
M5Gemma 4 E4B 4bitoMLX27.01800.0––10242026-10-13community–source
M5Gemma 4 E4B 4bitoMLX26.81559.0––10242026-10-11community–source
M5Gemma 4 E4B 4bitoMLX26.81559.0––10242026-10-11community–source
M5Gemma 4 E4B 4bitoMLX26.01760.0––40962026-10-11communitycontradictionsource
M5Gemma 4 E4B 4bitoMLX26.01760.0––40962026-10-11communitycontradictionsource
M4 MaxGemma 4 E4B oQ4eoMLX25.92298.0––40962026-10-11communitycontradiction outliersource
M5Gemma 4 E4B 4bitoMLX25.81161.0––10242026-10-13community–source
M5Gemma 4 E4B 4bitoMLX25.31736.0––40962026-10-13communitycontradictionsource
M5Gemma 4 E4B 4bitoMLX25.01988.0––40962026-10-13communitycontradictionsource
M5Gemma 4 E4B 4bitoMLX25.01988.0––40962026-10-13communitycontradictionsource
M5Gemma 4 E4B 4bitoMLX24.81979.0––40962026-10-13communitycontradictionsource
M4Gemma 4 E4B oQ4eoMLX24.1576.9––327682026-10-18community–source
M5Gemma 4 E4B 4bitoMLX23.41881.0––163842026-10-13communitycontradictionsource
M5Gemma 4 E4B 8bitoMLX22.91451.0––40962026-10-08community–source
M4Gemma 4 E4B 4bitoMLX22.9615.9––10242026-10-17communitycontradictionsource
M4Gemma 4 E4B 4bitoMLX22.6568.1––10242026-10-12communitycontradictionsource
M6Gemma 4 E4B 4bitoMLX22.31894.0––655362026-10-09community–source
M4Gemma 4 E4B 4bitoMLX22.2577.0––40962026-10-12communitycontradictionsource
M1 MaxGemma 4 E4B oQ4eoMLX22.2457.3––40962026-10-13community–source
M5Gemma 4 E4B 8bitoMLX20.71715.0––163842026-10-08community–source
M3Gemma 4 E4B 4bitoMLX20.2527.3––10242026-10-13community–source
M2 ProGemma 4 E4B oQ4eoMLX20.2603.2––10242026-10-15communitycontradictionsource
M3Gemma 4 E4B 4bitoMLX19.5546.2––40962026-10-13community–source
M4 Max (128 GB)Gemma 4 E4B Q8_0llama.cpp18.1–––327682026-10-17sourced–source
AMD Custom GPU 0932Gemma 4 E4B 7.52BQ4_K - Mediumllama.cpp (Vulkan)17.80163.93––tg1282026-08-20community–source
AMD Custom GPU 0932Gemma 4 E4B 7.52BQ4_K - Mediumllama.cpp (Vulkan)17.78161.37––tg1282026-08-20community–source
M3Gemma 4 E4B 8bitoMLX15.6519.1––10242026-10-10community–source
M3Gemma 4 E4B 8bitoMLX15.4542.5––40962026-10-10community–source
M4 Max (128 GB)Gemma 4 E4B Q8_0llama.cpp14.3–––1024002026-10-17sourced–source
Xeon 6985P-CGemma 4 E4B Q8_0llama.cpp (CPU)10.60–––tg1282026-10community–source
Xeon 6985P-CGemma 4 E4B Q8_0llama.cpp (CPU)10.43–––tg1282026-10community–source
Xeon 6985P-CGemma 4 E4B Q8_0llama.cpp (CPU)9.27–––tg1282026-10communitycontradictionsource
Xeon 6985P-CGemma 4 E4B Q8_0llama.cpp (CPU)9.16–––tg1282026-10communitycontradictionsource
Google Axion, Neoverse V2 with SVE2Gemma 4 E4B Q4_0llama.cpp (CPU)–126.69–––2026-10community–source
Google Axion, Neoverse V2 with SVE2Gemma 4 E4B Q4_0llama.cpp (CPU)–91.37–––2026-10community–source
Google Axion, Neoverse V2 with SVE2Gemma 4 E4B Q4_0llama.cpp (CPU)–247.07–––2026-10community–source
Google Axion, Neoverse V2 with SVE2Gemma 4 E4B Q4_0llama.cpp (CPU)–135.1–––2026-10community–source
Google Axion, Neoverse V2 with SVE2Gemma 4 E4B Q4_0llama.cpp (CPU)–121.03–––2026-10community–source
Google Axion, Neoverse V2 with SVE2Gemma 4 E4B Q4_0llama.cpp (CPU)–129.01–––2026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–19742.0––20482026-10-13community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–22494.2––20482026-10-13community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–23855.8––20482026-10-13community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–20097.0––40962026-10-13community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–22062.9––40962026-10-13community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–18361.2––10242026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–22787.7––10242026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–23480.9––10242026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–23924.9––10242026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–8918.4––5122026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–16890.1––5122026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–20515.7––5122026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–22436.7––5122026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–1666.5––1282026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–5943.5––1282026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–9302.1––1282026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–13435.8––1282026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–20474.2––40962026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–21364.6––40962026-10community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–22787.7–––2026-10-08community–source
A100-SXM4-80GBGemma 4 E4B BF16vLLM–34290.8–––2026-10-08community–source
Adreno X2-90Gemma 4 E4B Q4_K_Mllama.cpp (OpenCL)–565.2–––2026-10community–source
Adreno X2-90Gemma 4 E4B Q4_K_Mllama.cpp (OpenCL)–567.4–––2026-10community–source
Xeon 6985P-CGemma 4 E4B Q8_0llama.cpp (CPU)–152.06–––2026-10community–source
Xeon 6985P-CGemma 4 E4B Q8_0llama.cpp (CPU)–155.77–––2026-10community–source
Xeon 6985P-CGemma 4 E4B Q8_0llama.cpp (CPU)–39.75–––2026-10community–source
Xeon 6985P-CGemma 4 E4B Q8_0llama.cpp (CPU)–39.96–––2026-10community–source
Arc B580Gemma 4 E4B Q4_0llama.cpp (Vulkan)–896–––2026-10community–source
Arc B580Gemma 4 E4B Q4_0llama.cpp (Vulkan)–2731–––2026-10community–source
Adreno 840Gemma 4 E4B Q4_0llama.cpp–132––5122026-10-14sourced–source
Adreno 840Gemma 4 E4B Q4_0llama.cpp–152––5122026-10-14sourced–source

← all models