DeepSeek on Windows: Four Routes From Browser to API

If you have searched for “DeepSeek on Windows” expecting a single official installer, the answer is more nuanced. There are four legitimate routes, and which one fits depends on whether you want a chat window, an offline model running on your own GPU, or programmatic access to the DeepSeek API from PowerShell. This guide walks through each option with hardware requirements, exact commands, and the trade-offs of each. By the end you will know which route matches your machine, your privacy stance, and your budget — and which to avoid.
The short answer: pick one of four routes

DeepSeek does not ship a traditional “DeepSeek.exe” chat client for Windows. Instead, Windows users have four practical routes, and they cover almost every use case from casual chat to offline inference on a 128 GB workstation.
| Route | What you get | Best for | Cost |
|---|---|---|---|
| Browser at chat.deepseek.com | DeepSeek’s official web chat | Casual users, students, writers | Free |
| Web chat installed as an Edge app | The same web chat in its own window with a taskbar icon | Users who prefer a Start-menu icon | Free |
| Ollama + Open-WebUI | Local distilled or quantized weights, offline | Privacy-sensitive workflows, no internet | Hardware only |
| DeepSeek API from Python/PowerShell | Programmatic access to deepseek-flash and deepseek-v4-pro |
Developers, automation, agents | Pay per token |
On the API, deepseek-flash is served by DeepSeek-V4.1-Flash-1017 (released 17 October 2026) and deepseek-v4-pro by DeepSeek-V4-Pro-0813 (13 August 2026); both publish open weights under the MIT license. DeepSeek’s 17 October 2026 Change Log says the official web and app models were updated to V4.1-Flash-1017 as well. For the model family’s background, see DeepSeek V4.
Route 1: Use chat.deepseek.com in any Windows browser
This is the simplest route, and for most readers it is the right one. Open Edge, Chrome, or Firefox on Windows 10 or 11, go to chat.deepseek.com, and sign in. Nothing needs installing, and the chat runs the same model DeepSeek serves on its web and app surfaces.
What the web chat does that the API does not
The web chat keeps your conversation history server-side, so you can return to past chats from any device after signing in. The API, covered in Route 4, is stateless — the client must resend the full message history with every request. Do not assume one behaves like the other; this catches out a lot of first-time integrators. For a deeper walkthrough see our DeepSeek chat guide.
Route 2: Install the web chat as an Edge app
If you want DeepSeek to feel more like a desktop app without installing third-party software, Microsoft Edge can install any website as an app: Microsoft’s support page gives the path Settings and more (…) > More tools > Apps > Install this site as an app. The result opens the chat in its own window without the browser chrome, appears in the Start menu, and can be pinned to the taskbar or Start — while still being DeepSeek’s own web chat underneath.
- Open
chat.deepseek.comin Microsoft Edge and sign in. - Open Settings and more (…), then More tools > Apps > Install this site as an app.
- Confirm the name and click Install; pin the new window to the taskbar if you like.
Treat packaged “DeepSeek” apps with caution. When we searched the Microsoft Store catalogue (US listing) for “deepseek” on 23 October 2026, we found no app published by DeepSeek, and search results for “DeepSeek Windows download” surface third-party sites using the DeepSeek name that are not affiliated with the company — some bundle their own UI around the public API, others bundle questionable installers. If you want to verify what you are installing, our verify official DeepSeek app walkthrough lists the publisher fields and certificates to check.
Route 3: Run DeepSeek locally with Ollama on Windows 11

This is the route to pick if you want offline operation, no token billing, and conversations that never leave your machine. The catch is that you will not be running the current flagship weights at home — on Hugging Face the DeepSeek-V4-Pro-0813 repository holds about 893 GB of files and DeepSeek-V4.1-Flash-1017 about 510 GB (checked 23 October 2026). On consumer Windows hardware you will instead run a smaller R1 distillation.
Hardware reality check
As a rough rule of thumb for Ollama’s default 4-bit builds (not an official requirement; longer contexts need more memory):
- 8 GB VRAM (RTX 3060, 4060): the 7B R1 distill (a 4.7 GB download) fits.
- 16–24 GB VRAM (RTX 4080, 4090): the 14B (9.0 GB) and 32B (20 GB) distills fit, the 32B at the 24 GB end.
- 48 GB+ VRAM or unified memory: 70B distill (43 GB) territory.
- Full R1 (671B) or the current V4-family checkpoints: workstation or server only — hundreds of gigabytes of disk and a high-end multi-GPU rig.
For a deeper hardware breakdown see our DeepSeek system requirements guide.
Step-by-step install on Windows 11
- Install Ollama for Windows from
ollama.com. After installation it runs in the background as a tray application. - Open PowerShell and verify the install:
ollama -v. - Pull a model that fits your hardware. For a typical gaming PC:
ollama pull deepseek-r1:7b(a 4.7 GB download in Ollama’s library). - Run it:
ollama run deepseek-r1:7b. You now have a working chat in the terminal. - Optional — install Docker Desktop and enable WSL2, then run Open-WebUI in a container. Mapped to port 3000, it gives you a browser interface at
http://localhost:3000that lets you switch between any locally pulled models.
If you have an NVIDIA GPU, keep the driver current — Ollama’s Windows documentation asks for NVIDIA driver 551.61 or newer, and its GPU documentation lists support for compute capability 5.0 and newer — and Ollama will detect the card and offload layers to it automatically. Our running DeepSeek on Ollama tutorial goes deeper on Ollama itself, and the broader install DeepSeek locally tutorial covers the vLLM and Hugging Face route for server hardware.
Route 4: Call the DeepSeek API from Windows
If you write code, this is where DeepSeek is most interesting. Chat requests hit POST /chat/completions, the OpenAI-compatible endpoint at https://api.deepseek.com, and DeepSeek also exposes an Anthropic-compatible surface at https://api.deepseek.com/anthropic. That means the official OpenAI Python or Node SDK works against DeepSeek by changing only base_url and api_key.
Get a key, then call the endpoint
- Sign up at
platform.deepseek.comand create an API key. Our get a DeepSeek API key walkthrough covers the steps. - Install Python 3.10+ for Windows from the Microsoft Store or python.org.
- Open PowerShell and install the OpenAI SDK:
pip install openai. - Set your key as an environment variable so it is not hard-coded:
$env:DEEPSEEK_API_KEY = "sk-...".
A minimal Python call to deepseek-flash in non-thinking mode (thinking is the default, so it is switched off explicitly):
from openai import OpenAI
import os
client = OpenAI(
base_url="https://api.deepseek.com",
api_key=os.environ["DEEPSEEK_API_KEY"],
)
resp = client.chat.completions.create(
model="deepseek-flash",
messages=[{"role": "user", "content": "Summarise the V4 release."}],
temperature=1.3,
max_tokens=1024,
extra_body={"thinking": {"type": "disabled"}},
)
print(resp.choices[0].message.content)
To use thinking mode on either model, replace the extra_body line with extra_body={"thinking": {"type": "enabled"}} and add reasoning_effort="high" (or "low" / "max"); temperature has no effect in thinking mode. The response then returns reasoning_content alongside the final content. Output can run up to 384K tokens on both current models, so raise max_tokens for long reasoning at "max" effort.
Legacy IDs and migration
If you have older code referencing deepseek-chat or deepseek-reasoner, switch it now: those IDs were scheduled for retirement on 2026-07-24 at 15:59 UTC and are no longer listed in the official docs. The legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted, but they are served by DeepSeek-V4.1-Flash-1017. Migration is a one-line model= swap to deepseek-flash or deepseek-v4-pro; base_url does not change. See our DeepSeek OpenAI SDK compatibility notes for the full mapping.
What it costs to run on Windows
Pricing is per token, not per platform. As of October 2026, DeepSeek’s official pricing page lists, per 1M tokens, deepseek-flash at $0.006 cache-hit input, $0.30 cache-miss input and $1.20 output at peak ($0.003 / $0.15 / $0.60 off-peak), and deepseek-v4-pro at $0.044 / $1.32 / $3.96 at peak ($0.022 / $0.66 / $1.98 off-peak). Peak hours are 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday, excluding Chinese public holidays; every other hour is billed at the off-peak rate. Our DeepSeek pricing calculator and the DeepSeek API pricing guide handle the full math for your own workload.
Other Windows-side features worth knowing
- Context window: 1M tokens on both current models, with output up to 384K tokens.
- JSON mode: set
response_format={"type": "json_object"}. It is designed to return valid JSON, not guaranteed — include the word “json” plus a small example schema in your prompt, and setmax_tokenshigh enough to avoid truncation. - Streaming: set
stream=trueand you will receive server-sent events; when thinking is enabled the reasoning content streams alongside the final content. - Tool calling and FIM completion: tool calling works in both thinking and non-thinking modes; FIM (Fill-In-the-Middle) completion is Beta and non-thinking only — useful for IDE integrations on Windows.
- Temperature presets: in non-thinking mode, DeepSeek recommends 0.0 for code and maths, 1.0 for data analysis, 1.3 for general conversation and translation, 1.5 for creative writing.
Common Windows-specific issues

| Symptom | Likely cause | Fix |
|---|---|---|
| Ollama “model not found” | Mistyped tag (e.g. deepseek-r1:7 instead of :7b) |
Run ollama list to confirm exact tags |
| Ollama runs on CPU only | NVIDIA driver too old or GPU not supported | Update the driver (551.61 or newer), then quit and relaunch Ollama |
| Open-WebUI cannot reach Ollama | WSL2 networking | Bind Ollama to 0.0.0.0 and use host’s IP from WSL |
| API call returns 401 | Env var not set in the current PowerShell session | Re-set $env:DEEPSEEK_API_KEY or use setx for persistence |
| JSON mode returns empty content | Documented behaviour | Retry; ensure prompt contains “json” and a schema |
If your problems are broader than this, our DeepSeek troubleshooting guide covers login, app, and API errors across platforms, and the DeepSeek beginner guides hub indexes every platform-specific guide we maintain.
Which route should you actually pick?
- Pick the browser or the Edge-installed app if you want to chat, write, or research without managing infrastructure. This is the right answer for most readers.
- Pick Ollama if you need offline operation, your data cannot leave the machine, or you want to experiment with the open weights at a smaller scale.
- Pick the API if you are automating, building an internal tool, or running an agent. Start with
deepseek-flashfor most workloads; move todeepseek-v4-proonly when your own evaluation shows the lift is worth the price — it costs 3.3× as much on output and 4.4× on cache-miss input, at peak and off-peak alike.
Cross-platform users should also see DeepSeek on Mac for the equivalent setup on Apple Silicon.
Is there an official DeepSeek app for Windows?
Not as a native chat client that we could find: when we searched the Microsoft Store catalogue (US listing) on 23 October 2026, we found no app published by DeepSeek. The official surface on Windows is the web chat at chat.deepseek.com, which Edge can install as an app with its own window and Start-menu launcher. Avoid third-party “DeepSeek for Windows” installers found on download portals; they are not maintained by DeepSeek. See our DeepSeek download guide for verified sources.
How do I run DeepSeek offline on Windows 11?
Install Ollama from ollama.com, then run ollama pull deepseek-r1:7b in PowerShell to download a 7B distill that fits on an 8 GB GPU. For a graphical interface, layer Open-WebUI on top via Docker Desktop with WSL2. The full R1 (671B) and the current V4-family weights need a workstation or server. Our DeepSeek offline setup tutorial has the complete walkthrough.
Can my PC run DeepSeek V4.1-Flash locally?
Probably not without serious hardware. The DeepSeek-V4.1-Flash-1017 repository holds about 510 GB on Hugging Face, well past consumer GPU memory, and Ollama’s library lists V4.1-Flash only as cloud tags (checked 23 October 2026), which run on Ollama’s servers rather than your PC. Most Windows users with an RTX 4080 or 4090 should run a 14B–32B R1 distill instead, or use the API for the current models. Check our DeepSeek hardware calculator to size your machine.
What does the DeepSeek API cost from Windows?
Pricing is per token, not per platform. As of October 2026, deepseek-flash lists $0.30 cache-miss input and $1.20 output per 1M tokens at peak ($0.15 / $0.60 off-peak); deepseek-v4-pro lists $1.32 and $3.96 at peak ($0.66 / $1.98 off-peak). See our DeepSeek API pricing reference for worked examples.
Does DeepSeek on Windows still support the old deepseek-chat model ID?
Do not rely on it. The legacy IDs deepseek-chat and deepseek-reasoner were scheduled for retirement on 2026-07-24 at 15:59 UTC and are no longer listed in the official docs. Migrating is a one-line model= swap to deepseek-flash or deepseek-v4-pro; base_url does not change. Our DeepSeek API documentation hub covers the current model IDs.
Sources & Methodology
Primary sources
- OfficialDeepSeek API documentationEndpoints, request/response shape, model IDsLast checked: October 23, 2026
- PricingDeepSeek API pricing pagePer-1M-token rates, billing units, cache-hit pricingLast checked: October 23, 2026
- Model cardDeepSeek on Hugging FaceWeights, license, usage notes per modelLast checked: October 23, 2026
Technical references
- Technical reportDeepSeek-R1 paper (arXiv:2501.12948)R1 reasoning approach, distillation resultsLast checked: October 23, 2026
Context sources
- DocumentationDeepSeek API Change LogCurrent model versions (V4.1-Flash-1017, V4-Pro-0813), web/app model update of 17 Oct 2026, legacy-ID retirementLast checked: October 23, 2026
Methodology
Information in this article was checked against the official DeepSeek documentation, the article's cited primary sources, and the deepseekai.guide pricing dataset on the date shown above. Where third-party numbers are quoted, the originating source is linked inline so the reader can verify and re-check.
Data confidence
High for facts cited inline with primary-source links; medium for context narrative.
Editorial note
AI tooling moves quickly. If you spot an outdated number or a broken link, the contact link in the footer is the fastest way to flag it for the editorial team.
