Running
1
llm-bench.io
📈
Live leaderboard of local LLMs — real-world quality & speed
AI & ML interests
None defined yet.
Recent Activity
View all activity
Organization Card
llm-bench.io — Local LLM Leaderboard
Real-world benchmarks from actual users — LLM-judged quality and speed across coding, agent, role-play & research tasks, grouped by VRAM.
Run your own test →
| Model | VRAM | Runs | Quality | Speed (tok/s) | Coding | Agent | Role-Play | Research | Updated |
|---|---|---|---|---|---|---|---|---|---|
mtplx-qwen38-27b-optimized-quality
Apple M5 Max |
128 GB | 3 | 87.2 | 35.0 | 78.4 | 90.6 | 92.4 | 87.6 | 2026-09-01 |
unsloth/Qwen3.8-27B-GGUF:IQ3_S
NVIDIA GeForce RTX 4090 |
23 GB | 4 | 86.9 | 108.6 | 83.3 | 86.2 | 90.3 | 87.9 | 2026-08-29 |
Ornith-1.5-35B-A3B-oQ4e-mtp
Apple M5 Max |
64 GB | 3 | 85.8 | 112.0 | 82.8 | 84.9 | 89.8 | 85.6 | 2026-09-11 |
Dirk-Qwen3.8-27B-UD-IQ4_XS
NVIDIA GeForce RTX 5060 |
8 GB | 3 | 85.8 | 37.7 | 80.3 | 85.8 | 89.5 | 87.5 | 2026-09-11 |
mtplx-qwen38-27b-optimized-quality
Apple M4 Max |
64 GB | 3 | 85.7 | 35.5 | 77.7 | 85.2 | 93.1 | 86.7 | 2026-09-02 |
Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL
AMD Radeon RX 5600 OEM/5600 XT / 5700/5700 XT + AMD Radeon RX 9060 XT |
7 GB | 3 | 85.5 | 35.1 | 83.3 | 90.5 | 82.6 | 85.6 | 2026-09-04 |
unsloth/Qwen3.8-27B-GGUF:Q4_K_M
NVIDIA GeForce RTX 4090 |
23 GB | 6 | 85.1 | 95.3 | 78.0 | 85.9 | 89.0 | 87.4 | 2026-08-29 |
Qwen3.8-27B-oQ4e-fp16-mtp
Apple M5 Max |
128 GB | 4 | 85.1 | 38.6 | 82.5 | 85.1 | 89.6 | 83.0 | 2026-08-20 |
mtplx-flash-next-bare-speed
Apple M5 Max |
128 GB | 3 | 85.0 | 57.5 | 75.6 | 84.6 | 93.3 | 86.3 | 2026-09-01 |
qwen3.8-27b-UD-Q4_K_XL
AMD Radeon RX 7900 XT/7900 XTX/7900 GRE/7900M |
23 GB | 3 | 84.9 | 50.7 | 80.6 | 83.7 | 86.9 | 88.5 | 2026-09-08 |
Tiel-Coder-35B-A3B-Q4_K_S
AMD Radeon RX 7900 XT/7900 XTX/7900 GRE/7900M |
23 GB | 3 | 84.6 | 165.0 | 76.9 | 89.3 | 87.3 | 84.9 | 2026-09-05 |
Qwen3.8-27B-UD-Q3_K_XL
NVIDIA GeForce RTX 5070 Ti |
15 GB | 8 | 84.6 | 66.3 | 73.3 | 86.8 | 92.3 | 86.2 | 2026-09-04 |
Qwen3.8-27B-oQ8e-fp16-mtp
Apple M5 Max |
64 GB | 4 | 84.6 | 33.7 | 74.1 | 87.4 | 91.2 | 85.6 | 2026-08-28 |
Qwen3.8-27B-oQ4-mtp
Apple M5 Max |
64 GB | 4 | 83.9 | 42.8 | 75.4 | 89.1 | 85.0 | 86.1 | 2026-08-28 |
Muse-Glimmer-30B-oQ8e
Apple M5 Max |
64 GB | 3 | 83.9 | 17.0 | 72.3 | 89.4 | 92.0 | 82.1 | 2026-08-29 |
Qwen3.8-27B-GSQ-RCO-IQ3_S
NVIDIA GeForce RTX 4080 SUPER |
16 GB | 4 | 83.8 | 73.6 | 75.3 | 84.7 | 88.8 | 86.3 | 2026-09-04 |
Ornith-1.5-35B-A3B-oQ8e-mtp
Apple M5 Max |
64 GB | 5 | 83.7 | 94.6 | 77.5 | 82.0 | 88.5 | 86.8 | 2026-09-07 |
Qwen3.8-27B-oQ8e-mtp
Apple M4 Max |
128 GB | 5 | 83.7 | 31.1 | 78.6 | 84.9 | 84.8 | 86.4 | 2026-09-13 |
K2-Horizon-MoVA-36B-A4B-MLX-4bit
Apple M5 Max |
64 GB | 3 | 83.6 | 48.5 | 82.5 | 90.8 | 76.1 | 84.9 | 2026-09-11 |
Qwen3.8-27B-oQ4e-mtp
Apple M5 Max |
64 GB | 26 | 83.5 | 45.3 | 74.2 | 83.4 | 90.5 | 86.0 | 2026-09-09 |
qwen3.8:27b-mtp-q4_K_M
AMD Radeon RX 7900 XTX |
23 GB | 3 | 83.5 | 40.8 | 71.2 | 88.2 | 88.1 | 86.7 | 2026-08-16 |
Tiel-Coder-35B-A3B-MLX-oQ4e
Apple M5 Max |
64 GB | 4 | 83.4 | 118.4 | 80.7 | 82.7 | 85.6 | 84.4 | 2026-08-25 |
Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXS
NVIDIA GeForce RTX 5060 |
8 GB | 6 | 83.4 | 102.1 | 77.2 | 88.7 | 83.2 | 84.6 | 2026-09-11 |
Qwen3.8-27B-UD-IQ3_S
NVIDIA GeForce RTX 4080 SUPER |
16 GB | 48 | 83.2 | 78.8 | 77.8 | 81.6 | 87.8 | 85.7 | 2026-09-05 |
| 128 GB | 3 | 82.9 | 51.8 | 76.6 | 84.1 | 86.2 | 84.7 | 2026-09-12 | |
Qwen3.8-Flash-Next-oQ5e-mtp
Apple M4 Max |
128 GB | 64 | 82.8 | 56.3 | 74.0 | 84.0 | 89.3 | 83.9 | 2026-09-13 |
qwen3.8:27b
AMD Radeon RX 7900 XT/7900 XTX/7900 GRE/7900M |
23 GB | 5 | 82.8 | 35.6 | 78.6 | 82.8 | 84.8 | 84.9 | 2026-09-07 |
qwen3.8:latest
AMD Radeon RX 7900 XTX |
23 GB | 7 | 82.6 | 37.4 | 75.9 | 82.2 | 85.1 | 87.1 | 2026-09-07 |
Tiel-Coder-35B-A3B-MLX-oQ4e-MTP
Apple M5 Max |
64 GB | 5 | 82.1 | 123.7 | 76.3 | 84.4 | 82.3 | 85.3 | 2026-09-07 |
Qwen3.8-27B-4bit
Apple M5 Max |
64 GB | 9 | 81.9 | 30.6 | 76.8 | 78.0 | 89.2 | 83.4 | 2026-08-22 |
Qwen3.8-Flash-Next-oQ4e-mtp
Apple M4 Max |
128 GB | 50 | 81.8 | 54.2 | 72.3 | 83.1 | 87.4 | 84.4 | 2026-09-13 |
Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL
NVIDIA GeForce RTX 5060 |
8 GB | 4 | 81.7 | 108.1 | 77.3 | 80.5 | 83.7 | 85.4 | 2026-09-11 |
Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1
Apple M4 Max |
128 GB | 19 | 81.7 | 53.6 | 65.0 | 82.8 | 90.1 | 88.9 | 2026-09-13 |
| 128 GB | 3 | 81.7 | 48.7 | 69.0 | 86.2 | 85.1 | 86.6 | 2026-09-12 | |
Cyber-Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXS
NVIDIA GeForce RTX 5060 |
8 GB | 3 | 81.5 | 115.7 | 73.8 | 85.3 | 82.4 | 84.7 | 2026-09-10 |
Qwen3.8-27B-Q4_K_XL
AMD Radeon RX 7900 XT/7900 XTX/7900 GRE/7900M |
23 GB | 15 | 81.5 | 72.4 | 78.7 | 76.5 | 85.7 | 85.0 | 2026-09-05 |
| 128 GB | 3 | 81.3 | 59.5 | 69.3 | 84.2 | 91.5 | 80.0 | 2026-09-12 | |
Qwen3.8-27B-oQ8e-mtp
Apple M5 Max |
64 GB | 26 | 81.3 | 33.4 | 70.7 | 82.2 | 88.1 | 84.4 | 2026-09-10 |
| 128 GB | 4 | 81.1 | 63.8 | 70.3 | 85.2 | 87.0 | 82.0 | 2026-09-12 | |
unsloth/Qwen3.8-27B-GGUF:UD-Q8_K_XL
Apple M5 Max |
64 GB | 3 | 80.9 | 24.4 | 68.1 | 80.7 | 89.3 | 85.4 | 2026-08-21 |
Ornith-1.5-35B-A3B-IQ4_XS
AMD Radeon RX 7900 XT/7900 XTX/7900 GRE/7900M |
23 GB | 3 | 80.7 | 139.5 | 78.7 | 81.4 | 82.4 | 80.2 | 2026-09-05 |
Qwen3.8-27B-oQ8e-fp16-mtp
Apple M2 Ultra |
128 GB | 13 | 80.7 | 34.1 | 72.7 | 77.3 | 88.2 | 84.6 | 2026-09-04 |
Nex-N2.5-mini-MLX-4bit
Apple M5 Max |
64 GB | 20 | 80.3 | 137.0 | 73.9 | 81.2 | 81.1 | 85.1 | 2026-09-12 |
qwen3_coder_next
NVIDIA GeForce RTX 3080 |
10 GB | 4 | 80.1 | 31.2 | 72.5 | 87.3 | 85.3 | 75.4 | 2026-08-25 |
unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_XL
NVIDIA GeForce RTX 3080 Ti + NVIDIA GeForce RTX 3080 |
20 GB | 3 | 79.7 | 172.4 | 68.0 | 84.5 | 83.2 | 83.1 | 2026-09-07 |
qwen38-27b
AMD Radeon RX 7900 XT/7900 XTX/7900 GRE/7900M |
23 GB | 6 | 79.6 | 72.7 | 67.5 | 77.3 | 88.7 | 85.0 | 2026-09-08 |
Qwen3.8-27B-UD-Q6_K_XL
NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition |
96 GB | 4 | 79.5 | 84.3 | 75.9 | 76.9 | 84.4 | 80.6 | 2026-08-31 |
peculiar-ragdoll/Tiel-Coder-35B-A3B-GGUF-MTP
Advanced Micro Devices, Inc. [AMD/ATI] HawkPoint1 |
16 GB | 4 | 78.9 | 26.8 | 77.2 | 81.8 | 80.9 | 75.8 | 2026-09-01 |
Qwen3.8-27B-oQ8-mtp
Apple M5 Max |
64 GB | 4 | 77.7 | 35.6 | 77.6 | 67.4 | 84.8 | 81.0 | 2026-08-17 |
Qwen3.6-35B-A3B-IQ4_NL
AMD Radeon RX 7900 XT/7900 XTX/7900 GRE/7900M |
23 GB | 17 | 75.9 | 164.8 | 67.1 | 74.7 | 83.0 | 78.7 | 2026-09-05 |
gpt-oss-20b-UD-Q8_K_XL
NVIDIA GeForce RTX 5060 |
8 GB | 3 | 75.6 | 148.4 | 70.3 | 83.8 | 71.6 | 76.7 | 2026-09-09 |
gemma-4-26B-A4B-it-UD-IQ4_NL
NVIDIA GeForce RTX 5060 |
8 GB | 3 | 74.5 | 83.2 | 67.6 | 82.5 | 80.5 | 67.5 | 2026-09-12 |
gemma-4-26b-a4b-q4kxl
AMD Radeon RX 7900 XT/7900 XTX/7900 GRE/7900M |
23 GB | 3 | 69.5 | 169.9 | 67.9 | 75.2 | 79.8 | 55.2 | 2026-09-05 |
gemma4:26b-a4b-it-qat
AMD Radeon RX 7900 XTX |
23 GB | 3 | 69.2 | 116.0 | 60.5 | 81.7 | 70.1 | 64.3 | 2026-08-24 |
qwen3.5-9b-mtp@q8_0
AMD Radeon RX 7600 XT |
15 GB | 3 | 67.5 | 41.8 | 56.2 | 62.6 | 80.5 | 70.7 | 2026-09-11 |
qwen36-35b_VISION
NVIDIA GeForce RTX 3080 |
10 GB | 5 | 67.0 | 33.1 | 60.7 | 52.9 | 81.6 | 73.0 | 2026-08-26 |
mudler/Qwen3.6-35B-A3B-APEX-MTP-GGUF
Advanced Micro Devices, Inc. [AMD/ATI] HawkPoint1 |
16 GB | 4 | 66.8 | 31.6 | 65.6 | 59.9 | 68.0 | 73.8 | 2026-09-01 |
qwen35_122b
NVIDIA GeForce RTX 3080 |
10 GB | 5 | 66.7 | 15.2 | 53.7 | 56.3 | 80.3 | 76.3 | 2026-08-25 |
Ling-3.0-tiny-oQ8e
Apple M4 Max |
128 GB | 4 | 60.1 | 124.9 | 47.3 | 72.3 | 45.5 | 75.2 | 2026-09-10 |
ornith-1.5-9b
NVIDIA GeForce RTX 3080 |
10 GB | 3 | 54.3 | 93.8 | 11.7 | 55.2 | 76.6 | 73.8 | 2026-08-25 |
ling-oq4e
Apple M1 |
16 GB | 4 | 46.1 | 47.6 | 6.2 | 64.1 | 59.5 | 54.8 | 2026-09-12 |
ling-oq8e
Apple M1 |
16 GB | 6 | 44.0 | 33.8 | 6.0 | 73.7 | 45.3 | 51.1 | 2026-09-12 |
Qwen3.8-27B-oQ2e-mtp
Apple M5 Max |
64 GB | 3 | 10.3 | 44.0 | 0.0 | 28.0 | 7.7 | 5.6 | 2026-08-30 |
models 0
None public yet