Bonsai 2 27B
23 records across 5 hardware.
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Arc B580 | Bonsai 2 27B | PTQ1_0 | llama.cpp (SYCL) | 250 | – | – | – | 131072 | 2026-09 | community | – | source |
| RTX 5060 Ti | Bonsai 2 27B | PQ2_0 | llama.cpp | 127.8 | – | – | – | 32768 | 2026-10-10 | community | – | source |
| RTX 5060 Ti | Bonsai 2 27B | PQ2_0 | llama.cpp | 127.3 | – | – | – | 131072 | 2026-10-10 | community | – | source |
| RTX 5060 Ti | Bonsai 2 27B | PQ2_0 | llama.cpp | 114.8 | – | – | – | 245760 | 2026-10-10 | community | – | source |
| RTX 5060 Ti | Bonsai 2 27B | PQ2_0 | llama.cpp | 112.4 | – | – | – | 196608 | 2026-10-10 | community | – | source |
| RTX 5060 Ti | Bonsai 2 27B | PQ2_0 | llama.cpp | 111.8 | – | – | – | 229376 | 2026-10-10 | community | – | source |
| Arc B580 | Bonsai 2 27B | PTQ1_0 | llama.cpp (SYCL) | 90 | – | – | – | 131072 | 2026-09 | community | – | source |
| RTX 4080 SUPER | Bonsai 2 27B | 5.95 GB | llama.cpp (CUDA) | 85 | – | – | – | 262144 | 2026-10-08 | community | – | source |
| RTX 5060 Ti | Bonsai 2 27B | PQ2_0 | llama.cpp | 79.7 | – | – | – | 262144 | 2026-10-10 | community | – | source |
| RTX 5060 Ti | Bonsai 2 27B | PQ2_0 | llama.cpp | 47.9 | – | – | – | 262144 | 2026-10-10 | community | – | source |
| Arc B580 | Bonsai 2 27B | PTQ1_0 | llama.cpp (SYCL) | 44 | – | – | – | 131072 | 2026-09 | community | – | source |
| Arc B580 | Bonsai 2 27B | PTQ1_0 | llama.cpp (SYCL) | 42 | 895 | – | – | 131072 | 2026-09 | community | – | source |
| Arc B580 | Bonsai 2 27B | PTQ1_0 | llama.cpp (SYCL) | 36 | – | – | – | 32768 | 2026-09 | community | – | source |
| Arc B580 | Bonsai 2 27B | PTQ1_0 | llama.cpp (SYCL) | 26.4 | – | – | – | 131072 | 2026-09 | community | – | source |
| M5 Pro | Bonsai 2 27B | unknown | oMLX | 22.4 | 248.1 | – | – | 1024 | 2026-10-10 | community | – | source |
| M5 Pro | Bonsai 2 27B | unknown | oMLX | 21.3 | 289.3 | – | – | 4096 | 2026-10-10 | community | – | source |
| Arc B580 | Bonsai 2 27B | PTQ1_0 | llama.cpp (PrismML SYCL fork) | 21 | 207 | – | – | 131072 | 2026-09 | community | – | source |
| RTX 5060 Ti | Bonsai 2 27B | PQ2_0 | llama.cpp | 13.3 | – | – | – | 262144 | 2026-10-10 | community | – | source |
| RTX 5060 Ti | Bonsai 2 27B | PQ2_0 | llama.cpp | 11.0 | – | – | – | 524288 | 2026-10-10 | community | – | source |
| Radeon 860M | Bonsai 2 27B | Q2_0 | llama.cpp (Vulkan) | 8.17 | 66.4 | – | – | tg64 | 2026-09-30 | community | – | source |
| Arc B580 | Bonsai 2 27B | PTQ1_0 | llama.cpp (PrismML SYCL fork) | 8 | – | – | – | 32768 | 2026-09 | community | – | source |
| Radeon 860M | Bonsai 2 27B | Q2_0 | llama.cpp (Vulkan) | 7.26 | 59.2 | – | – | tg64 | 2026-09-30 | community | – | source |
| RTX 4080 SUPER | Bonsai 2 27B | 5.95 GB | llama.cpp (CUDA) | – | 2278 | – | – | 262144 | 2026-10-08 | community | – | source |