TeichAI / deepseek-v3 family
deepseek-v3.2-speciale
deepseek/deepseek-v3.2-speciale
Run locally Smallest indexed GGUF / needs Ollama
ollama run hf.co/deepseek/deepseek-v3.2-speciale/smol-IQ1_KT/DeepSeek-V3.2-Speciale-smol-IQ1_KT-00001-of-00005.gguf Benchmarks
Where it scores.
General Provisional
58.6
Coding Provisional
69.6
Math Provisional
66.3
Reasoning Provisional
60.3
Normalized 0–100 across indexed models.
Runnable variants
Pick a size.
| Precision | Format | Size | Params | File |
|---|---|---|---|---|
| IQ1_KT / smallest | GGUF | 156.9 GB | — | smol-IQ1_KT/DeepSeek-V3.2-Speciale-smol-IQ1_KT-00001-of-00005.gguf |
| IQ2_KS | GGUF | 192.7 GB | — | smol-IQ2_KS/DeepSeek-V3.2-Speciale-smol-IQ2_KS-00001-of-00005.gguf |
| IQ3_K | GGUF | 312.4 GB | — | IQ3_K/DeepSeek-V3.2-Speciale-IQ3_K-00001-of-00008.gguf |
| IQ5_K | GGUF | 498.7 GB | — | IQ5_K/DeepSeek-V3.2-Speciale-IQ5_K-00001-of-00012.gguf |
Sizes are file sizes, not peak runtime memory.
Raw benchmarks
The inputs.
| Source | Task | Raw | Normalized |
|---|---|---|---|
| ZeroEval | AIME2025 | 96.0 | 96.0 |
| ZeroEval | HLE | 30.6 | 30.6 |
| ZeroEval | SWEBenchVerified | 73.1 | 73.1 |
From the Hugging Face model card
About this model.
From the Hugging Face model card