MiMo-V2.6-Flash-RL

70 records across 9 hardware.

hardwaremodelquantbackendtok/spp tok/sWttft sctx/tgdateclassflagssource
4x AMD MI210MiMo-V2.6-Flash-RL 309Bmxfp4vLLM160––––2026-09-29community–source
M5 UltraMiMo-V2.6-Flash-RL 4bitoMLX104.83207.0––40962026-10-08community–source
M5 UltraMiMo-V2.6-Flash-RL 4bitoMLX92.93381.0––81922026-10-08community–source
M5 UltraMiMo-V2.6-Flash-RL 4bitoMLX84.91844.0––10242026-10-08community–source
M5 UltraMiMo-V2.6-Flash-RL mxfp4oMLX71.4552.0––10242026-10-07community–source
M5 UltraMiMo-V2.6-Flash-RL 4bitoMLX68.62329.0––327682026-10-08community–source
4x CMP 170HXMiMo-V2.6-Flash-RL FP8vLLM63–––51202026-09-28community–source
4x CMP 170HXMiMo-V2.6-Flash-RL FP8vLLM56.1–––670002026-09-28community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX55.5427.9––10242026-10-07community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX54.9636.6––40962026-10-12community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX54.5484.4––40962026-10-07community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX54.3621.0––81922026-10-12community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX53.6447.2––163842026-10-16community–source
M3 UltraMiMo-V2.6-Flash-RL mxfp4oMLX53.3617.7––81922026-10-10community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX53.3482.2––40962026-10-16community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX53.1424.0––10242026-10-16community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX52.8473.1––81922026-10-07community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX51.7409.1––10242026-10-07community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX51.7673.1––40962026-10-10community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX51.5480.7––40962026-10-07community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX51.2531.9––10242026-10-12community–source
M3 UltraMiMo-V2.6-Flash-RL mxfp4oMLX51.0585.5––163842026-10-10community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX50.9606.1––10242026-10-10community–source
M2 UltraMiMo-V2.6-Flash-RL 4bitoMLX50.2344.0––10242026-10-17community–source
M2 UltraMiMo-V2.6-Flash-RL 4bitoMLX50.2344.0––10242026-10-17community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX50.1450.1––163842026-10-07community–source
M3 UltraMiMo-V2.6-Flash-RL mxfp4oMLX50.1632.5––40962026-10-10community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX49.4644.3––81922026-10-16community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX48.7655.5––81922026-10-10communitycontradictionsource
M2 UltraMiMo-V2.6-Flash-RL 4bitoMLX48.5375.5––40962026-10-17community–source
M2 UltraMiMo-V2.6-Flash-RL 4bitoMLX48.5375.5––40962026-10-17community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX47.7471.5––81922026-10-07community–source
M5 MaxMiMo-V2.6-Flash-RL mxfp4oMLX47.41293.0––10242026-10-14community–source
M5 MaxMiMo-V2.6-Flash-RL mxfp4oMLX47.4579.6––10242026-10-13community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX47.1587.4––163842026-10-12community–source
M5 MaxMiMo-V2.6-Flash-RL mxfp4oMLX46.41300.0––40962026-10-13community–source
M3 UltraMiMo-V2.6-Flash-RL mxfp4oMLX46.2522.0––327682026-10-10community–source
M5 MaxMiMo-V2.6-Flash-RL mxfp4oMLX46.01753.0––81922026-10-13community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX46.0607.0––163842026-10-16community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX45.6618.3––163842026-10-10community–source
M5 MaxMiMo-V2.6-Flash-RL mxfp4oMLX44.91461.0––163842026-10-13community–source
M3 UltraMiMo-V2.6-Flash-RL oQ8eoMLX44.7573.2––10242026-10-14community–source
M3 UltraMiMo-V2.6-Flash-RL oQ8eoMLX44.7573.2––10242026-10-14community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX44.2402.8––327682026-10-07community–source
M3 UltraMiMo-V2.6-Flash-RL oQ8eoMLX44.1644.6––40962026-10-14community–source
M3 UltraMiMo-V2.6-Flash-RL oQ8eoMLX44.1644.6––40962026-10-14community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX43.9646.2––40962026-10-16communitycontradictionsource
M3 UltraMiMo-V2.6-Flash-RL mxfp4oMLX43.5545.0––10242026-10-10community–source
4x CMP 170HXMiMo-V2.6-Flash-RL FP8vLLM42.5–––51202026-09-28community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX41.4547.6––327682026-10-10community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX41.3447.7––163842026-10-07communitycontradictionsource
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX40.9389.3––327682026-10-07community–source
M2 UltraMiMo-V2.6-Flash-RL 4bitoMLX40.7325.8––327682026-10-18community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX37.1330.3––655362026-10-07community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX34.4439.8––655362026-10-10community–source
4x CMP 170HXMiMo-V2.6-Flash-RL FP8vLLM33.7–––670002026-09-28community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX33.0304.0––655362026-10-07communitycontradictionsource
M3 UltraMiMo-V2.6-Flash-RL mxfp4oMLX31.4424.2––655362026-10-10community–source
4x AMD MI210MiMo-V2.6-Flash-RL 309Bmxfp4vLLM31––––2026-09-29community–source
M3 UltraMiMo-V2.6-Flash-RL mxfp4oMLX28.6243.7––1310722026-10-10community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX25.4309.8––1310722026-10-10community–source
Radeon 8065SMiMo-V2.6-Flash-RL llama-bench -fa 1 -b 2048 -ub 2048, 3 runs; ROCmFPX main (fused gate/up expert tensors, from PR #33), Mesa RADVMXFP4-Q5K mixed (4.29 bpw)llama.cpp (Vulkan)23.6–––tg1282026-10-02community–source
Radeon 8065SMiMo-V2.6-Flash-RL llama-bench -fa 1 -b 2048 -ub 2048, 3 runs; ROCmFPX main (fused gate/up expert tensors, from PR #33), Mesa RADVMXFP4-Q5K mixed (4.29 bpw)llama.cpp (Vulkan)22.1–––327682026-10-02community–source
M3 UltraMiMo-V2.6-Flash-RL 4bitoMLX20.5233.1––1996802026-10-10community–source
M3 UltraMiMo-V2.6-Flash-RL mxfp4oMLX19.1220.3––1996802026-10-10community–source
8× NVIDIA RTX 3080 20GBMiMo-V2.6-Flash-RL MXFP4llama.cpp18.6–––2621442026-09community–source
4x CMP 170HXMiMo-V2.6-Flash-RL FP8vLLM16.5–––670002026-09-28community–source
Radeon 8060SMiMo-V2.6-Flash-RL 309BIQ2llama.cpp (Vulkan)2.6–––327682026-10-05community–source
Radeon 8065SMiMo-V2.6-Flash-RL llama-bench -fa 1 -b 2048 -ub 2048, 3 runs; ROCmFPX main (fused gate/up expert tensors, from PR #33), Mesa RADVMXFP4-Q5K mixed (4.29 bpw)llama.cpp (Vulkan)–484.9–––2026-10-02community–source
Radeon 8065SMiMo-V2.6-Flash-RL llama-bench -fa 1 -b 2048 -ub 2048, 3 runs; ROCmFPX main (fused gate/up expert tensors, from PR #33), Mesa RADVMXFP4-Q5K mixed (4.29 bpw)llama.cpp (Vulkan)–258.7––327682026-10-02community–source

← all models