Best local LLM for coding
The models people report running for coding on their own hardware, ranked from community reports. Pick your memory size first: a model that is the best choice on a 24 GB card is often not the best on 12 GB.
Ranked from reports whose posters used the model for coding.
By VRAM
All memory sizes
| # | Model | Reports | Gen t/s | PP t/s | With speculation |
|---|---|---|---|---|---|
| 1 | Qwen3.827B Alibaba · mostly Q4_K_M Gen 31 t/s (36 runs) · PP 803 t/s (12 runs) · with speculation 65 t/s (69) | 215 | 31 36 runs, 18 devices | 803 12 runs | 65 69 runs |
| 2 | Qwen3.8 Flash-Next125B · 6B active Alibaba · mostly IQ3_S Gen 54 t/s (24 runs) · PP 930 t/s (14 runs) · with speculation 44 t/s (18) | 171 | 54 24 runs, 9 devices | 930 14 runs | 44 18 runs |
| 3 | Qwen3.635B · 3B active Alibaba · mostly Q4_K_M Gen 57 t/s (16 runs) · PP 298 t/s (6 runs) · with speculation 114 t/s (6) | 47 | 57 16 runs, 11 devices | 298 6 runs | 114 6 runs |
| 4 | Qwen3.627B Alibaba · mostly Q4_K_M Gen 25 t/s (7 runs) · PP 706 t/s (3 runs) · with speculation 50 t/s (12) | 36 | 25 7 runs, 5 devices | 706 3 runs | 50 12 runs |
| 5 | DeepSeek V4 Flash284B · 13B active DeepSeek · mostly IQ2_M Gen 120 t/s (2 runs) · with speculation 17 t/s (2) | 23 | 120 2 runs | no PP data | 17 2 runs |
| 6 | DeepSeek V4.1 Flash552B · 16B active mostly MXFP4 Gen 16 t/s (1 run) · PP 300 t/s (1 run) · with speculation 31 t/s (1) | 17 | 16 1 run | 300 1 run | 31 1 run |
| 7 | Qwen3.8 Swift-1.527B Alibaba · mostly MXFP4 Gen 53 t/s (3 runs) · with speculation 86 t/s (6) | 16 | 53 3 runs, 3 devices | no PP data | 86 6 runs |
| 8 | GLM-5.3 Flash320B · 18B active Zhipu AI · mostly NVFP4 Gen 37 t/s (3 runs) · PP 875 t/s (2 runs) | 15 | 37 3 runs, 2 devices | 875 2 runs | none |
| 9 | Ornith1.535B · 3B active mostly Q4_K_M Gen 69 t/s (1 run) · with speculation 162 t/s (2) | 9 | 69 1 run | no PP data | 162 2 runs |
| 10 | Qwen3-Coder30B · 3B active mostly Q4_K_M Gen 122 t/s (4 runs) | 7 | 122 4 runs, 3 devices | no PP data | none |
| 11 | Bonsai 227B PrismML · mostly PTQ1_0 Gen 50 t/s (6 runs) · PP 1053 t/s (4 runs) | 6 | 50 6 runs, 5 devices | 1053 4 runs | none |
| 12 | Muse Glimmer30B Meta · mostly K-Quant-Dynamic No generation data · with speculation 129 t/s (4) | 6 | no generation data | no PP data | 129 4 runs |
| 13 | Gemma 426B · 4B active Google DeepMind · mostly Q5_K_M Gen 129 t/s (3 runs) · PP 5600 t/s (1 run) | 4 | 129 3 runs, 2 devices | 5600 1 run | none |
| 14 | Gemma 431B Google DeepMind · mostly NVFP4 Gen 37 t/s (1 run) · with speculation 96 t/s (2) | 4 | 37 1 run | no PP data | 96 2 runs |
| 15 | Qwen3.8 Swift-1.5-125B-A6B125B · 6B active Alibaba · mostly EXL3 4.05 Gen 36 t/s (1 run) · PP 1634 t/s (1 run) | 4 | 36 1 run | 1634 1 run | none |
| 16 | Qwen3-Coder-Next80B · 3B active mostly Q4_K_M Gen 24 t/s (3 runs) · PP 546 t/s (1 run) | 4 | 24 3 runs | 546 1 run | none |
| 17 | Qwen3.535B · 3B active Alibaba · mostly IQ4_NL Gen 50 t/s (1 run) · PP 980 t/s (1 run) | 3 | 50 1 run | 980 1 run | none |
| 18 | DeepSeek V4 Flash Vision-Exp285B · 13B active DeepSeek · mostly NVFP4 No generation data · with speculation 32 t/s (1) | 3 | no generation data | no PP data | 32 1 run |
| 19 | Qwen3.8 Swift27B Alibaba · mostly BF16 No generation data | 3 | no generation data | no PP data | none |
| 20 | Qwen3.8 Uncensored27B Alibaba · mostly 4-bit No generation data · with speculation 92 t/s (3) | 3 | no generation data | no PP data | 92 3 runs |
| 21 | Qwen2.5 Coder-7B7B Alibaba · mostly Q4_K_M Gen 159 t/s (1 run) | 2 | 159 1 run | no PP data | none |
| 22 | Qwen38B Alibaba · mostly Q4 Gen 87 t/s (2 runs) | 2 | 87 2 runs, 2 devices | no PP data | none |
| 23 | Nemotron 3.5 Bolt74B · 9B active mostly 2-bit Gen 85 t/s (1 run) · PP 833 t/s (1 run) | 2 | 85 1 run | 833 1 run | none |
| 24 | Qwen3.8 Escha-W227B Alibaba · mostly 2.469 bpw Gen 63 t/s (2 runs) | 2 | 63 2 runs, 2 devices | no PP data | none |
| 25 | GLM-4.7 Flash30B Zhipu AI · mostly Q4_K_XL Gen 43 t/s (1 run) | 2 | 43 1 run | no PP data | none |
| 26 | Gemma 412B Google DeepMind · mostly Q5_K_XL Gen 39 t/s (2 runs) | 2 | 39 2 runs | no PP data | none |
| 27 | KAT-Coder V2.5-Dev35B · 3B active Kwaipilot · mostly W4A16 GPTQ Gen 14 t/s (1 run) | 2 | 14 1 run | no PP data | none |
| 28 | Bonsai 2 Ternary27B PrismML · mostly 2-bit ternary No generation data · with speculation 188 t/s (1) | 2 | no generation data | no PP data | 188 1 run |
| 29 | DeepSeek V4.1 Flash763B mostly GSQ No generation data | 2 | no generation data | no PP data | none |
| 30 | Kimi K32800B · 104B active Moonshot AI No generation data | 2 | no generation data | no PP data | none |
| 31 | Mellum2.112B · 2.5B active mostly Q8 No generation data | 2 | no generation data | no PP data | none |
| 32 | Nemotron 3.5 Lightning30B · 3B active NVIDIA · mostly Q4_0 No generation data | 2 | no generation data | no PP data | none |
| 33 | Qwen3.59B Alibaba · mostly 4-bit No generation data · with speculation 198 t/s (1) | 2 | no generation data | no PP data | 198 1 run |
| 34 | Qwen3.8 Swift-1.5-Uncensored27B Alibaba · mostly 2.3bpw No generation data · with speculation 68 t/s (1) | 2 | no generation data | no PP data | 68 1 run |
| 35 | Qwen3.8 Huihui-Abliterated27B Alibaba · mostly NVFP4 No generation data · with speculation 203 t/s (2) | 2 | no generation data | no PP data | 203 2 runs |
| 36 | Qwen3.8 Swift-1.5 Flash-Next125B · 6B active Alibaba · mostly IQ2_XS No generation data · with speculation 145 t/s (1) | 2 | no generation data | no PP data | 145 1 run |
| 37 | Qwen3.8 Flash-Next Uncensored125B · 6B active Alibaba · mostly NVFP4 No generation data | 2 | no generation data | no PP data | none |
| 38 | Nex-N2.5-mini35B · 3B active mostly MLX-4bit Gen 134 t/s (1 run) | 1 | 134 1 run | no PP data | none |
| 39 | DeepSeek-Coder-V2-Lite16B · 2.4B active mostly 4bit Gen 126 t/s (1 run) · PP 364 t/s (1 run) | 1 | 126 1 run | 364 1 run | none |
| 40 | Qwen2.5 Coder-1.5B1.5B Alibaba Gen 99 t/s (1 run) | 1 | 99 1 run | no PP data | none |
| 41 | MiniCPM-V 4.735B · 3B active mostly mix_q3gud_im Gen 93 t/s (1 run) | 1 | 93 1 run | no PP data | none |
| 42 | gpt-oss117B · 5.1B active OpenAI · mostly MXFP4 Gen 56 t/s (1 run) · PP 1135 t/s (1 run) | 1 | 56 1 run | 1135 1 run | none |
| 43 | Qwen3.5 reasoning9B Alibaba Gen 50 t/s (1 run) | 1 | 50 1 run | no PP data | none |
| 44 | DeepSeek V4 Flash 9B distill9B DeepSeek · mostly Q4_K_M Gen 44 t/s (1 run) | 1 | 44 1 run | no PP data | none |
| 45 | Mimo 2.6 Distill-Qwen9B Xiaomi · mostly Q8_0 Gen 42 t/s (1 run) · PP 1718 t/s (1 run) | 1 | 42 1 run | 1718 1 run | none |
| 46 | Qwen3.6 APEX-I-Quality35B · 3B active Alibaba Gen 37 t/s (1 run) | 1 | 37 1 run | no PP data | none |
| 47 | Qwen3.8 AREX-227B Alibaba · mostly EXL3 4.00 bpw Gen 36 t/s (1 run) | 1 | 36 1 run | no PP data | none |
| 48 | Qwen2.57B Alibaba Gen 34 t/s (1 run) | 1 | 34 1 run | no PP data | none |
| 49 | Nemotron 3.5 Super VL120B · 12B active mostly Q4_K_M Gen 31 t/s (1 run) | 1 | 31 1 run | no PP data | none |
| 50 | Nanbeige4.333B mostly 4-bit Gen 29 t/s (1 run) · PP 757 t/s (1 run) | 1 | 29 1 run | 757 1 run | none |
| 51 | Qwen2.5 Coder-32B32B Alibaba · mostly Q4_K_M Gen 28 t/s (1 run) · PP 450 t/s (1 run) | 1 | 28 1 run | 450 1 run | none |
| 52 | Minimax M2.1230B · 10B active mostly 4bit Gen 25 t/s (1 run) | 1 | 25 1 run | no PP data | none |
| 53 | Agents-A1 Millie35B · 3B active InternScience · mostly ternary Gen 22 t/s (1 run) · PP 261 t/s (1 run) | 1 | 22 1 run | 261 1 run | none |
| 54 | Qwen3-Next80B · 3B active Alibaba · mostly Q4_K_M Gen 19 t/s (1 run) · PP 130 t/s (1 run) | 1 | 19 1 run | 130 1 run | none |
| 55 | Llama 3.21B Meta · mostly Q4_K_M Gen 15 t/s (1 run) · PP 108 t/s (1 run) | 1 | 15 1 run | 108 1 run | none |
| 56 | Qwen3.6 Coder27B · 3B active Alibaba · mostly Q4_K_M Gen 14 t/s (1 run) | 1 | 14 1 run | no PP data | none |
| 57 | Ling-3.0 Tiny7.9B · 1.3B active Ant Group · mostly Q6 Gen 10 t/s (1 run) | 1 | 10 1 run | no PP data | none |
| 58 | Gemma 4 E4B8B Google DeepMind · mostly 4bit Gen 8.2 t/s (1 run) | 1 | 8.2 1 run | no PP data | none |
| 59 | Qwen 2.5 Coder14B mostly Q3_K_M Gen 1.2 t/s (1 run) | 1 | 1.2 1 run | no PP data | none |
| 60 | Bonsai 2 Swift Katana27B PrismML · mostly PTQ1_0 No generation data | 1 | no generation data | no PP data | none |
| 61 | DeepSeek V4 Flash REAP-150B150B DeepSeek · mostly FP4 No generation data | 1 | no generation data | no PP data | none |
| 62 | Gemma 426B Google DeepMind No generation data | 1 | no generation data | no PP data | none |
| 63 | Mimo 2.5310B · 15B active Xiaomi No generation data | 1 | no generation data | no PP data | none |
| 64 | Mimo 2.6 Flash309B · 15B active Xiaomi No generation data | 1 | no generation data | no PP data | none |
| 65 | Mimo 2.6 Flash-RL309B · 15B active Xiaomi · mostly MXFP4 No generation data | 1 | no generation data | no PP data | none |
| 66 | Nemotron 3.5 Bolt74B · 9B active mostly IQ2_XXS No generation data · with speculation 343 t/s (1) | 1 | no generation data | no PP data | 343 1 run |
| 67 | Ornith1.59B mostly Q6_K No generation data | 1 | no generation data | no PP data | none |
| 68 | Ornith1.5 BigBang35B · 3B active mostly Q4_K_M No generation data · with speculation 169 t/s (1) | 1 | no generation data | no PP data | 169 1 run |
| 69 | Ornith1.5122B mostly UD-IQ2_XXS No generation data | 1 | no generation data | no PP data | none |
| 70 | Qwen2.532B Alibaba · mostly AWQ No generation data | 1 | no generation data | no PP data | none |
| 71 | Qwen2.5-Coder3B Alibaba No generation data | 1 | no generation data | no PP data | none |
| 72 | Qwen314B Alibaba · mostly Q4_K_M No generation data | 1 | no generation data | no PP data | none |
| 73 | Qwen330B · 3B active Alibaba · mostly Q4_K_M No generation data | 1 | no generation data | no PP data | none |
| 74 | Qwen3.54B Alibaba · mostly Q4_K_M No generation data | 1 | no generation data | no PP data | none |
| 75 | Qwen3.5 Mica v0.14B Alibaba · mostly Q4_K_M No generation data | 1 | no generation data | no PP data | none |
| 76 | Qwen3.5 DeepSeek-V4-Flash9B Alibaba · mostly Q6_K No generation data · with speculation 49 t/s (1) | 1 | no generation data | no PP data | 49 1 run |
| 77 | Qwen3.527B Alibaba · mostly Q4_K_M No generation data | 1 | no generation data | no PP data | none |
| 78 | Qwen3.5122B · 10B active Alibaba · mostly UD-IQ4_XS No generation data | 1 | no generation data | no PP data | none |
| 79 | Qwen3.6 Fable-Fus-711-UnHeretic-NM-DAU-NEO-MAX-NEO-LOW27B Alibaba · mostly IQ4_XS No generation data · with speculation 42 t/s (1) | 1 | no generation data | no PP data | 42 1 run |
| 80 | Qwen3.6 Tiel-Coder35B · 3B active Alibaba · mostly Q4_K_XL No generation data | 1 | no generation data | no PP data | none |
| 81 | Qwen3.8 Splash-HQ27B Alibaba · mostly Q8 No generation data · with speculation 37 t/s (1) | 1 | no generation data | no PP data | 37 1 run |
| 82 | Qwen3.8 L0xRE27B Alibaba · mostly IQ3 No generation data · with speculation 40 t/s (1) | 1 | no generation data | no PP data | 40 1 run |
| 83 | Qwen3.8 QUASAR27B Alibaba · mostly NVFP4 No generation data · with speculation 287 t/s (1) | 1 | no generation data | no PP data | 287 1 run |
| 84 | Qwen3.8 CODER27B Alibaba · mostly IQ4_XS No generation data | 1 | no generation data | no PP data | none |
| 85 | Qwen3.8 Cold-Fusion-GAIN-V1.1-NM-DAU-NEO-MAX-NEO27B Alibaba · mostly Q8_0 No generation data | 1 | no generation data | no PP data | none |
| 86 | Qwen3.8 abliterated27B Alibaba · mostly IQ3_S No generation data · with speculation 105 t/s (1) | 1 | no generation data | no PP data | 105 1 run |
| 87 | Qwen3.8 pi27B Alibaba · mostly Q4_K_M No generation data | 1 | no generation data | no PP data | none |
| 88 | Qwen3.8 ASCII-Condensed27B Alibaba · mostly UD-IQ4_XS No generation data | 1 | no generation data | no PP data | none |
| 89 | Qwen3.8 RVN27B Alibaba · mostly Q6_K No generation data · with speculation 100 t/s (1) | 1 | no generation data | no PP data | 100 1 run |
| 90 | Qwen3.8 RVN Heretic (ARA abliterated)27B Alibaba · mostly Q6_K No generation data | 1 | no generation data | no PP data | none |
| 91 | Qwen3.835B Alibaba · mostly Q6 No generation data | 1 | no generation data | no PP data | none |
| 92 | Qwopus3.6 Coder-Compat27B mostly Q4_K_M No generation data · with speculation 40 t/s (1) | 1 | no generation data | no PP data | 40 1 run |
Frequently asked
What is the best local LLM for coding?
It depends on how much memory you have, which is why this page is split by VRAM. Each tier lists the model sizes people report using for coding on that much memory, with the quant most of them used and their typical speed, so you can pick from setups like yours rather than from a single global list.
How are these rankings built?
From community performance reports tagged as coding use. Model sizes are ordered by how many reports use them. Typical speed is the median of plain runs on one device, and runs with speculative decoding are shown separately. Nothing is sponsored.
Does a coding model need more VRAM than a chat model?
Not for the weights, but coding sessions carry long contexts: whole files, diffs and tool output. The KV cache for that context has to fit next to the weights, so leave more headroom than a chat use would need, or pick a quant one rung smaller.