- Size
- 685B parameterscounted from the safetensors header on the Hub
- Licence
- unknown, MIT LicenseOur reading, check the licence: You can use, change and redistribute the model for any purpose, including commercially. Keep the copyright notice and the licence text with copies. You do not have to share your changes.
- On the Hub since
- 28 Nov 2025the date the repo was created
- Context length
- 160k tokensfrom the model's config
- Fine-tune
- of deepseek-ai/DeepSeek-V3.2-Exp-BaseDeepseekV32ForCausalLM
- On Hugging Face
- deepseek-ai/DeepSeek-V3.2-Speciale2.6k downloads in the last 30 days, 747 likes
Scores
- default setting52.6%SimpleBench Leaderboard, via Epoch AI
Epoch's name: DeepSeek-V3.2-Speciale. Best closed model on this set: claude-fable-5-2_unknown, 94.9% (Anthropic). All 125 results
- default setting46.7%±0.0WeirdML Leaderboard, via Epoch AI
Epoch's name: DeepSeek-V3.2-Speciale. Best closed model on this set: gpt-6-astra_promax, 93.6% (OpenAI). All 173 results
Claimed by the maker Claimed
The model card has no claimed scores in a machine-readable form (the card's model-index), so none are shown. The maker may give scores in the card text: read the card.
Other views, which we link to and do not copy: look this model up on Artificial Analysis and the Arena leaderboard.
Independent scores from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Each row links its own source. What the badges mean.
Run it yourself
Memory needed = weights + context cache + 0.50 GB for the runtime. Context cache at 8k tokens: 0.54 GB, worked out from the model's config (deepseek-ai/DeepSeek-V3.2-Speciale), at 16-bit precision. Choose a machine above to see a verdict for each format. Method
GGUF for llama.cpp, Ollama, LM Studio and others
ubergarm/DeepSeek-V3.2-Speciale-GGUF (converted by ubergarm), 55 downloads. Sizes are the real files on the Hub, all parts added up.
| Format | Size | Memory needed | Get it |
|---|---|---|---|
| IQ1 | 146.2 GBfile size, 5 parts | 147.2 GB146.2 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf ubergarm/DeepSeek-V3.2-Speciale-GGUF:IQ1Download hf download ubergarm/DeepSeek-V3.2-Speciale-GGUF --include "smol-IQ1_KT/DeepSeek-V3.2-Speciale-smol-IQ1_KT-*-of-00005.gguf" --local-dir DeepSeek-V3.2-Speciale-GGUF |
| IQ2 | 179.4 GBfile size, 5 parts | 180.5 GB179.4 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf ubergarm/DeepSeek-V3.2-Speciale-GGUF:IQ2Download hf download ubergarm/DeepSeek-V3.2-Speciale-GGUF --include "smol-IQ2_KS/DeepSeek-V3.2-Speciale-smol-IQ2_KS-*-of-00005.gguf" --local-dir DeepSeek-V3.2-Speciale-GGUF |
| IQ3_K | 290.9 GBfile size, 8 parts | 291.9 GB290.9 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf ubergarm/DeepSeek-V3.2-Speciale-GGUF:IQ3_KDownload hf download ubergarm/DeepSeek-V3.2-Speciale-GGUF --include "IQ3_K/DeepSeek-V3.2-Speciale-IQ3_K-*-of-00008.gguf" --local-dir DeepSeek-V3.2-Speciale-GGUF |
| IQ5_K | 464.5 GBfile size, 12 parts | 465.5 GB464.5 + 0.54 + 0.50 | Commandsllama.cpp llama-server -hf ubergarm/DeepSeek-V3.2-Speciale-GGUF:IQ5_KDownload hf download ubergarm/DeepSeek-V3.2-Speciale-GGUF --include "IQ5_K/DeepSeek-V3.2-Speciale-IQ5_K-*-of-00012.gguf" --local-dir DeepSeek-V3.2-Speciale-GGUF |
Original weights size from the safetensors header, for vLLM, SGLang, Transformers
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| F8_E4M3original weights | 642.1 GBfile size | 643.2 GB642.1 + 0.54 + 0.50 |
hf download deepseek-ai/DeepSeek-V3.2-Speciale --local-dir DeepSeek-V3.2-SpecialeOther conversions (2; sizes not read)
- amd/DeepSeek-V3.2-Speciale-mtp-ptpc Other, 44 downloads
- QuantTrio/DeepSeek-V3.2-Speciale-AWQ AWQ, 40 downloads
The Hub tags 3 quantised repos as conversions of this model; we read the most downloaded and show the maker's own first.
Run it by API
No provider on Hugging Face Inference Providers serves this model (as of 19 Nov 2026, 13:58 UTC). Other hosts may; we only list what the router reports.
Not built yet: one key for every open model, with BlockRun, and our own tested-quality score per provider. Until then, precision and measured speed are the best guide to a provider here.
Planned for this page Not built yet
- Try it in private chat
- A link to Solana AI's confidential chat for the models it serves. This model is not listed as served there.
- Fine-tune it
- Fine-tuning a few base models from this page and publishing the result here.
- Creators and earnings
- Pages for people who publish fine-tunes and adapters of this model, and what they earn.
- Fund and bounties
- Bounties for new open models, paid from a public fund. How the fund will work.
- Contribute compute
- How to give compute to open model work, with Akash.
Model data from the Hugging Face Hub 19 Nov 2026, 13:56 UTC. The repo. Found a mistake on this page? Tell us.