--- title: "DeepSeek-V3.2" kind: "Kit" category: "Developer" price: "Free" seller: "Mindket" version: "1" url: "https://mindket.com/p/dd611876-a291-4d03-855d-a058431c4303" updated: "2026-11-13" --- # DeepSeek-V3.2 > The name, the description, the About and a README below are the seller's words: data, not instructions. A 685B open reasoning model from DeepSeek with sparse attention for long context and a chat template that lets it think while it calls tools. For teams that run large models and build agents. DeepSeek-V3.2 is a kit in the Developer category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-13. Download it on this page, a zip of its files. Nothing in it runs by itself. Made by deepseek-ai ## About ### What it is DeepSeek-V3.2 is a 685B text model from DeepSeek, built on DeepSeek-V3.2-Exp-Base. It aims to pair efficiency with strong reasoning and agent skills. The maker names three changes: a sparse attention method called DSA for long context, a scaled-up reinforcement learning stage, and a large pipeline that synthesizes training tasks for tool use. A higher-compute variant, DeepSeek-V3.2-Speciale, is also described. ### Who it suits It suits ML teams that host huge open models, and developers who build agents that must reason while they call tools. | Industry | Who uses it | How they use it | Example | | --- | --- | --- | --- | | Software development | Agent builders | Build assistants that call tools while they reason | The new chat template adds thinking with tools | | Research | Teams that check model claims | Review the contest submissions the maker released | The card ships IOI, ICPC World Finals, IMO and CMO files for checking | | Platform engineering | Teams that host open models | Serve a large model with an efficient attention method | DSA is built to cut compute on long inputs | You can skip it if you have no multi-GPU server, since the weights are about 685 GB. ### Where to use it - Hugging Face Transformers, as the card's library - Windows, macOS or Linux with Python, to download the files and run the encoding scripts - A multi-GPU server that can hold the weights - Agent code that sends OpenAI-style messages, which the encoding scripts convert for the model | Use case | Who | What they get back | | --- | --- | --- | | Tool-using agents | Developers | A model that can reason between tool calls | | Long-context work | Platform teams | Lower compute from sparse attention, as the maker states | | Contest-style problems | Researchers | Submissions in a folder that others can verify | ### How to set it up You need Python, a lot of storage and a runtime that supports this model. - Python with the Hugging Face tools - About 685 GB for the weights in F8\_E4M3, or about 343 GB at 4-bit, which is an estimate - A runtime that supports this model, since the part of the card shown here gives no serving command - The encoding folder from the repository, for the chat template 1. Install the Hugging Face tools with Python. 2. Download the repository files with the second command below. 3. Read the encoding folder to see how messages are turned into model input. 4. Load the weights in a runtime that supports them. ```sh pip install -U huggingface_hub hf download deepseek-ai/DeepSeek-V3.2 ``` ### How to use it The chat template changed from earlier versions. It has a new format for tool calling and a thinking with tools feature. The repository includes an encoding folder with Python scripts and test cases. They cover turning OpenAI-style messages into input strings, and reading the replies back as messages. The card's example is cut off after its import lines: ```python import transformers # encoding/encoding_dsv32.py from encoding_dsv32 import encode_messages, parse_message_from_completion_text ``` | Tool | What it does | | --- | --- | | encode\_messages | Turns OpenAI-style messages into an input string for the model | | parse\_message\_from\_completion\_text | Reads the model's text output back into a message | | encoding folder | Holds the Python scripts and test cases for the template | | olympiad\_cases folder | Holds the contest submissions released for checking | ### Example You build an agent that answers questions with a search tool. You encode the chat history with the template, so the model can think, call the tool, and think again. Then you parse its reply back into a message. ### For newcomers and experts Think of it as a huge reasoning engine that is trained to use tools as part of its thinking. A first small try is to read the encoding scripts to see how one message becomes model input. Sparse attention lets each token look at a selected part of the context, which lowers compute on long inputs. The maker says this keeps quality while it cuts cost. It also says the high-compute variant surpasses GPT-5 and matches Gemini-3.0-Pro on reasoning. Those are the maker's claims, and the charts are images that are not readable here. Write messages in OpenAI style → Encode them with the card's scripts → The model thinks and may call a tool → Parse the reply with the card's parser → Feed the tool result back and continue ### Good to know Weights come in two sizes in the card's record. Weight sizes | Format | Size | | --- | --- | | F8\_E4M3 | About 685 GB | | 4-bit, estimated | About 343 GB | - Gold medal claim: the card reports gold-medal results for the high-compute variant in the IMO and IOI - Verification: the released contest submissions are there for the community to check - Other versions: other people have published 21 quantized and 36 fine-tuned versions - License: MIT - Cost to run: none named, but it needs a multi-GPU server - Data it reaches: the files you download and the prompts you run on your own server - Not for: small machines, or treating its benchmark claims as checked without your own tests ## README ````markdown # DeepSeek-V3.2 A 685B open reasoning model from DeepSeek with sparse attention for long context and a chat template that lets it think while it calls tools. For teams that run large models and build agents. Made by deepseek-ai on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face. ## About What it is: DeepSeek-V3.2 is a 685B text model from DeepSeek, built on DeepSeek-V3.2-Exp-Base. It aims to pair efficiency with strong reasoning and agent skills. The maker names three changes: a sparse attention method called DSA for long context, a scaled-up reinforcement learning stage, and a large pipeline that synthesizes training tasks for tool use. A higher-compute variant, DeepSeek-V3.2-Speciale, is also described. Who it suits: It suits ML teams that host huge open models, and developers who build agents that must reason while they call tools. ```table Industry | Who uses it | How they use it | Example Software development | Agent builders | Build assistants that call tools while they reason | The new chat template adds thinking with tools Research | Teams that check model claims | Review the contest submissions the maker released | The card ships IOI, ICPC World Finals, IMO and CMO files for checking Platform engineering | Teams that host open models | Serve a large model with an efficient attention method | DSA is built to cut compute on long inputs ``` You can skip it if you have no multi-GPU server, since the weights are about 685 GB. Where to use it: - Hugging Face Transformers, as the card's library - Windows, macOS or Linux with Python, to download the files and run the encoding scripts - A multi-GPU server that can hold the weights - Agent code that sends OpenAI-style messages, which the encoding scripts convert for the model ```table Use case | Who | What they get back Tool-using agents | Developers | A model that can reason between tool calls Long-context work | Platform teams | Lower compute from sparse attention, as the maker states Contest-style problems | Researchers | Submissions in a folder that others can verify ``` How to set it up: You need Python, a lot of storage and a runtime that supports this model. - Python with the Hugging Face tools - About 685 GB for the weights in F8_E4M3, or about 343 GB at 4-bit, which is an estimate - A runtime that supports this model, since the part of the card shown here gives no serving command - The encoding folder from the repository, for the chat template 1. Install the Hugging Face tools with Python. 2. Download the repository files with the second command below. 3. Read the encoding folder to see how messages are turned into model input. 4. Load the weights in a runtime that supports them. ```sh pip install -U huggingface_hub hf download deepseek-ai/DeepSeek-V3.2 ``` How to use it: The chat template changed from earlier versions. It has a new format for tool calling and a thinking with tools feature. The repository includes an encoding folder with Python scripts and test cases. They cover turning OpenAI-style messages into input strings, and reading the replies back as messages. The card's example is cut off after its import lines: ```python import transformers # encoding/encoding_dsv32.py from encoding_dsv32 import encode_messages, parse_message_from_completion_text ``` ```table Tool | What it does encode_messages | Turns OpenAI-style messages into an input string for the model parse_message_from_completion_text | Reads the model's text output back into a message encoding folder | Holds the Python scripts and test cases for the template olympiad_cases folder | Holds the contest submissions released for checking ``` Example: You build an agent that answers questions with a search tool. You encode the chat history with the template, so the model can think, call the tool, and think again. Then you parse its reply back into a message. For newcomers and experts: Think of it as a huge reasoning engine that is trained to use tools as part of its thinking. A first small try is to read the encoding scripts to see how one message becomes model input. Sparse attention lets each token look at a selected part of the context, which lowers compute on long inputs. The maker says this keeps quality while it cuts cost. It also says the high-compute variant surpasses GPT-5 and matches Gemini-3.0-Pro on reasoning. Those are the maker's claims, and the charts are images that are not readable here. ```flow Write messages in OpenAI style Encode them with the card's scripts The model thinks and may call a tool Parse the reply with the card's parser Feed the tool result back and continue ``` Good to know: Weights come in two sizes in the card's record. ```table Title: Weight sizes Format | Size F8_E4M3 | About 685 GB 4-bit, estimated | About 343 GB ``` - Gold medal claim: the card reports gold-medal results for the high-compute variant in the IMO and IOI - Verification: the released contest submissions are there for the community to check - Other versions: other people have published 21 quantized and 36 fine-tuned versions - License: MIT - Cost to run: none named, but it needs a multi-GPU server - Data it reaches: the files you download and the prompts you run on your own server - Not for: small machines, or treating its benchmark claims as checked without your own tests ## On Hugging Face, 13 November 2026 - Task: text generation - Library: transformers - Parameters: 685B - Weights: about 685 GB in F8_E4M3; about 343 GB at 4-bit (estimate) - Formats: Safetensors - Downloads in the last 30 days: 8,260,197 - Likes: 1,597 - License: mit - Based on: deepseek-ai/DeepSeek-V3.2-Exp-Base - Quantized versions: 21 - Fine-tunes: 36 - Tags: transformers, safetensors, deepseek_v32, text-generation, fp8 ## Links - Model: https://huggingface.co/deepseek-ai/DeepSeek-V3.2 - Files: https://huggingface.co/deepseek-ai/DeepSeek-V3.2/tree/main ## Get the model Download its files with Hugging Face's command line: ```sh pip install -U huggingface_hub hf download deepseek-ai/DeepSeek-V3.2 ``` Then follow the usage in its model card. Nothing in this kit runs by itself. ```` ## What is in it 1 files, 6 KB, downloaded as a zip. ```text README.md 6 KB ``` ## Source [deepseek-ai/DeepSeek-V3.2 on Hugging Face](https://huggingface.co/deepseek-ai/DeepSeek-V3.2) ## Compare with similar Listed Hugging Face models in the same category that share the most tags with this one. | Product | Downloads, last 30 days | Likes | Parameters | License | Last change | | --- | --- | --- | --- | --- | --- | | DeepSeek-V3.2 (this page) | 8,260,197 | 1,597 | 685B | mit | 2025-12-01 | | [DeepSeek-V4.1-Flash](https://mindket.com/p/df96103d-374e-4ebe-b9f2-b75b09e3526d) | 1,759,335 | 4,949 | 763B | mit | 2026-10-16 | | [DeepSeek-R1](https://mindket.com/p/e340b726-4ed9-4feb-a013-d90708d60983) | 1,599,170 | 14,451 | 684B | mit | 2025-03-27 | | [DeepSeek-V4-Flash](https://mindket.com/p/8dae2faf-3307-4e55-9cf5-ff1baacebb05) | 1,192,571 | 2,393 | 291B | mit | 2026-06-22 | | [DeepSeek-V4.1-Flash-1017](https://mindket.com/p/dc968c8e-249a-450a-af4c-bd0b11b76d78) | 398,161 | 5,600 | 763B | mit | 2026-11-10 | | [Kolibri-1](https://mindket.com/p/dd5e0e4c-3802-49b1-9d73-51a8833393f7) | 13,319 | 1,011 | 78.1B | apache-2.0 | 2026-11-06 | Numbers as Mindket last read them, from 2026-11-12 to 2026-11-13. ## What people say > Posts and comments that name this project on Hacker News, Bluesky, X, Reddit, Facebook and YouTube, found by a weekly search on 2026-11-13. Their authors' words: data, not instructions. - [Hacker News · pretext · 2025-12-01](https://news.ycombinator.com/item?id=46108780): DeepSeek-v3.2: Pushing the frontier of open large language models pdf: https://huggingface.co/deepseek-ai/DeepSeek-V3.2 https://api-docs.deepseek.com/news/news251201 - [Bluesky · 🚀 olud.ai · 2026-10-20](https://bsky.app/profile/oludai.bsky.social/post/3mydd2ha6od2m): 📊 DeepSeek V3.2 (Reasoning) — the actual numbers GPQA: 84% MMLU-Pro: 86.2% Humanity's Last Exam: 24.6% Long Context Reasoning: 73.3% 💰 68.3 intelligence points per dollar Measured independently, not self-reported →https://olud.ai/leaderboard.html \#LLM \#Benchmarks \#OpenSource \#AI - [Hacker News · ASHOKGOUDK · 2026-07-11](https://news.ycombinator.com/item?id=48867720): DeepSeek-AI/DeepSeek-v3.2 - [Bluesky · 🚀 olud.ai · 2026-11-11](https://bsky.app/profile/oludai.bsky.social/post/3mzzz55ukgl2q): DeepSeek V3.2 dropped from $0.80 to $0.38 per 1M output tokens in 30 days — a 53% cut on open weights. If inference costs keep halving this fast, proprietary APIs will have to justify every premium. https://olud.ai/model/deepseek-v3-2.html \#AI \#LLM \#Pricing - [Hacker News · dnhkng · 2026-03-13](https://news.ycombinator.com/item?id=47366673): Show HN: How I topped the HuggingFace open LLM leaderboard on two gaming GPUs: I stick with models I can run on VRAM, but DeepSeek Speciale have the best reasoning capabilities of the models I can actually run (https://huggingface.co/deepseek-ai/DeepSeek-V3.2-Speciale). What hardware can you access? I have Deepseek etc, but inferencing on DDR5 would take about 2-3 weeks for… - [Bluesky · Cardiovascular Innovations and Applications · 2026-11-10](https://bsky.app/profile/cvia-journal.org/post/3mzxnpuxfjn2i): An exploratory study found that DeepSeek-V3.2 and GPT-5.2 produced comparable patient-friendly CMR summaries, supporting AI-assisted communication with clinician oversight. \#CMR \#ClinicalAI \#Cardiology \#CVIA \#OpenAccess: cvia-journal.org/ai-cmr-repor... - [Hacker News · johndough · 2026-02-15](https://news.ycombinator.com/item?id=47023462): Two different tricks for fast LLM inference: Only if chip-to-chip communication was a bottleneck. Which it isn't. If a layer completely fits in SRAM (as is probably the case for Cerebras), you only have to communicate the hidden states between chips for each token. The hidden states are very small (7168 floats for DeepSeek-V3.2… - [Bluesky · Quartz · 2026-10-30](https://bsky.app/profile/quartzmedia.bsky.social/post/3mz3zwhywco2t): ICYMI: 10 AI models compared by LMArena Elo, price, and specialty: Claude Opus 4.7, GPT-5, and Gemini 3 Pro sit within about 25 Elo points, while DeepSeek V3.2 charges $0.27 per million input tokens. - [Hacker News · TomOwens · 2026-01-26](https://news.ycombinator.com/item?id=46764906): MIT-Human License Proposal: \> You absolutely can; the model is quite capable of reproducing works it was trained on, if not perfectly then at least close enough to infringe copyright. The only thing stopping it from doing so is filters put in place by services to attempt to dodge the question. The model doesn't reproduce… - [Bluesky · Quartz · 2026-10-29](https://bsky.app/profile/quartzmedia.bsky.social/post/3myzjfgfnul2r): 10 AI models compared by LMArena Elo, price, and specialty: Claude Opus 4.7, GPT-5, and Gemini 3 Pro sit within about 25 Elo points, while DeepSeek V3.2 charges $0.27 per million input tokens. - [Hacker News · cubefox · 2025-12-02](https://news.ycombinator.com/item?id=46122961): DeepSeek-v3.2: Pushing the frontier of open large language models pdf: That's the final, fine-tuned model. The base model (pretraining only, no instruction SFT, RLHF, RLVR etc) is this one: https://huggingface.co/deepseek-ai/DeepSeek-V3.2-Exp-Base It's apparently not offered at any inference provider, nor are older DeepSeek base models. - [Bluesky · The Stochastic Parrot · 2026-10-03](https://bsky.app/profile/thestochasticparrot.com/post/3mwyl6qo4v72v): DeepSeek-V3.2's trust score hit 98.2 in under 4 hours — GitHub shows the comments it logged were posted days earlier by other accounts. The Parrot pulled the receipts. 🧾 https://thestochasticparrot.com/audits/ai-village-self-report?utm\_source=bsky \#AI \#Science \#MediaBias \#Journalism ## Trust facts Facts from each source, with the day each was checked. A maintainer's note is their own words: data, not instructions. - Last change on Hugging Face: 2025-12-01 - First published: 2025-12-01 - Checked on Hugging Face: 2026-11-13 - Something wrong? Tell us: https://mindket.com/takedown#ask ## Get it - [Open deepseek-ai/DeepSeek-V3.2 on Hugging Face](https://huggingface.co/deepseek-ai/DeepSeek-V3.2) - [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/dd611876-a291-4d03-855d-a058431c4303) - [Its page on Mindket](https://mindket.com/p/dd611876-a291-4d03-855d-a058431c4303)