unsloth / deepseek-v3 family
deepseek-v3.1
deepseek/deepseek-v3.1
Run locally Smallest indexed GGUF / needs Ollama
ollama run hf.co/deepseek/deepseek-v3.1/DeepSeek-V3.1-UD-TQ1_0.gguf Benchmarks
Where it scores.
General
74.0
Coding
59.8
Math
55.6
Reasoning
51.0
Normalized 0–100 across indexed models.
Runnable variants
Pick a size.
| Precision | Format | Size | Params | File |
|---|---|---|---|---|
| Q4 / smallest | GGUF | 170.5 GB | — | DeepSeek-V3.1-UD-TQ1_0.gguf |
| IQ1_S | GGUF | 192.3 GB | — | UD-IQ1_S/DeepSeek-V3.1-UD-IQ1_S-00001-of-00004.gguf |
| IQ1_M | GGUF | 206.8 GB | — | UD-IQ1_M/DeepSeek-V3.1-UD-IQ1_M-00001-of-00005.gguf |
| IQ2_XXS | GGUF | 225.9 GB | — | UD-IQ2_XXS/DeepSeek-V3.1-UD-IQ2_XXS-00001-of-00005.gguf |
| IQ2_M | GGUF | 235.5 GB | — | UD-IQ2_M/DeepSeek-V3.1-UD-IQ2_M-00001-of-00005.gguf |
| Q2_K | GGUF | 245.7 GB | — | Q2_K/DeepSeek-V3.1-Q2_K-00001-of-00005.gguf |
| Q2_K_L | GGUF | 245.9 GB | — | Q2_K_L/DeepSeek-V3.1-Q2_K_L-00001-of-00005.gguf |
| Q2_K_XL | GGUF | 255.7 GB | — | UD-Q2_K_XL/DeepSeek-V3.1-UD-Q2_K_XL-00001-of-00006.gguf |
| IQ3_XXS | GGUF | 279.7 GB | — | UD-IQ3_XXS/DeepSeek-V3.1-UD-IQ3_XXS-00001-of-00006.gguf |
| Q3_K_S | GGUF | 290.4 GB | — | Q3_K_S/DeepSeek-V3.1-Q3_K_S-00001-of-00006.gguf |
| Q3_K_XL | GGUF | 300.0 GB | — | UD-Q3_K_XL/DeepSeek-V3.1-UD-Q3_K_XL-00001-of-00007.gguf |
| Q3_K_M | GGUF | 320.5 GB | — | Q3_K_M/DeepSeek-V3.1-Q3_K_M-00001-of-00007.gguf |
| IQ4_XS | GGUF | 358.3 GB | — | IQ4_XS/DeepSeek-V3.1-IQ4_XS-00001-of-00008.gguf |
| IQ4_NL | GGUF | 379.1 GB | — | IQ4_NL/DeepSeek-V3.1-IQ4_NL-00001-of-00008.gguf |
| Q4_0 | GGUF | 380.1 GB | — | Q4_0/DeepSeek-V3.1-Q4_0-00001-of-00008.gguf |
| Q4_K_S | GGUF | 381.1 GB | — | Q4_K_S/DeepSeek-V3.1-Q4_K_S-00001-of-00008.gguf |
| Q4_K_XL | GGUF | 386.9 GB | — | UD-Q4_K_XL/DeepSeek-V3.1-UD-Q4_K_XL-00001-of-00008.gguf |
| Q4_K_M | GGUF | 405.4 GB | — | Q4_K_M/DeepSeek-V3.1-Q4_K_M-00001-of-00009.gguf |
| Q4_1 | GGUF | 420.8 GB | — | Q4_1/DeepSeek-V3.1-Q4_1-00001-of-00009.gguf |
| Q5_K_S | GGUF | 462.6 GB | — | Q5_K_S/DeepSeek-V3.1-Q5_K_S-00001-of-00010.gguf |
| Q5_K_M | GGUF | 476.2 GB | — | Q5_K_M/DeepSeek-V3.1-Q5_K_M-00001-of-00010.gguf |
| Q5_K_XL | GGUF | 484.6 GB | — | UD-Q5_K_XL/DeepSeek-V3.1-UD-Q5_K_XL-00001-of-00010.gguf |
| Q6_K | GGUF | 551.3 GB | — | Q6_K/DeepSeek-V3.1-Q6_K-00001-of-00012.gguf |
| Q6_K_XL | GGUF | 574.5 GB | — | UD-Q6_K_XL/DeepSeek-V3.1-UD-Q6_K_XL-00001-of-00012.gguf |
| Q8_0 | GGUF | 713.3 GB | — | Q8_0/DeepSeek-V3.1-Q8_0-00001-of-00015.gguf |
| Q8_K_XL | GGUF | 780.6 GB | — | UD-Q8_K_XL/DeepSeek-V3.1-UD-Q8_K_XL-00001-of-00017.gguf |
| BF16 | GGUF | 1342.3 GB | — | BF16/DeepSeek-V3.1-BF16-00001-of-00030.gguf |
Sizes are file sizes, not peak runtime memory.
Raw benchmarks
The inputs.
| Source | Task | Raw | Normalized |
|---|---|---|---|
| Arena | ArenaScore | 1416.9 | 89.6 |
| ZeroEval | AIME2025 | 49.8 | 49.8 |
| ZeroEval | GPQA | 74.9 | 74.9 |
| ZeroEval | HLE | 15.9 | 15.9 |
| ZeroEval | SWEBenchVerified | 66.0 | 66.0 |
| ZeroEval | SimpleQA | 93.4 | 93.4 |
From the Hugging Face model card
About this model.
From the Hugging Face model card