How to Use DeepSeek on Mac: Web, App, Local and API

If you own a MacBook Air, a Mac mini, or a Mac Studio and want to use DeepSeek without juggling a Linux box, you have four working paths: the web chat in Safari, DeepSeek’s official apps, a locally hosted open-weight model through Ollama or LM Studio, and the developer API. DeepSeek’s current models — DeepSeek-V4.1-Flash-1017 and DeepSeek-V4-Pro-0813 — are published as open weights under the MIT license, but they are far larger than typical Mac memory, so local use on most Macs means smaller distilled models. This guide walks through each option, including realistic hardware expectations for local inference, sample API code, current pricing, and the trade-offs nobody mentions until you hit them.
The four ways to run DeepSeek on a Mac

Before you install anything, pick the right access path. Each one optimises for a different goal.
| Method | Best for | Cost | Privacy | Setup time |
|---|---|---|---|---|
| Web chat (chat.deepseek.com) | Casual use, quick questions | Free | Servers in China | 2 minutes |
| Official apps (iPhone/iPad chat app; DeepSeek Harness desktop preview) | Chat on mobile; agent tasks on an Apple silicon Mac | Chat app free; Harness is open source and bills through your DeepSeek API key | Servers in China | 5 minutes |
| Local via Ollama / LM Studio | Offline work, privacy, tinkering | Hardware only | Stays on device | 20–60 minutes |
API (deepseek-flash / deepseek-v4-pro) |
Building apps, scripts, agents | Pay per token | Servers in China | 10 minutes |
The right choice depends on three questions: do you need offline access, do you handle sensitive data, and do you want to write code against it. If you answered no to all three, the web chat is the easiest start.
Option 1: DeepSeek web chat in Safari, Chrome or Arc
The simplest route is chat.deepseek.com in any modern browser on macOS. Sign in with your email or phone number and password, or with Google or Apple where available. According to DeepSeek’s 17 October 2026 Change Log, the official web and app models were updated to DeepSeek-V4.1-Flash-1017, and DeepSeek’s homepage shows DeepThink and Search options next to its chat link. There is no install, no Rosetta concern, no Apple Silicon vs Intel question — it runs wherever Safari runs.
Things to know about the web surface:
- Conversation history is preserved server-side per account, unlike the API.
- You can upload files and photos; DeepSeek’s privacy policy lists both among the inputs it collects.
- DeepThink has the model reason before it answers, which takes longer.
- For an account walkthrough, see the sign up for DeepSeek guide.
If you mostly chat and occasionally paste code, the web is enough. Move to a heavier option only when you have a reason to.
Option 2: The official DeepSeek app on Apple Silicon
DeepSeek’s download page (checked 23 October 2026) offers two kinds of app. The DeepSeek chat app — “Chat with leading AI models for free” — is offered for iOS and Android; its US App Store listing gives compatibility for iPhone, iPad and iPod touch and does not list Mac. If the app does show up in the Mac App Store for you, confirm the seller is Hangzhou DeepSeek Artificial Intelligence Co., Ltd before installing — there have been counterfeits. The verify official DeepSeek app guide covers the spoofing risk in detail.
The same page offers DeepSeek Harness for desktop, with a “Download for macOS” build for Apple silicon on macOS 13 or later (a Windows build is also listed). Harness is DeepSeek’s open-source agent app, in public preview: it works in the background, edits local files and handles longer tasks. Its quickstart has you enter a DeepSeek API key under Settings → Models (other providers and custom OpenAI-compatible endpoints are also supported), so usage is billed at API rates; server-side search and fetch calls made through Harness were free of charge up to 00:00 Beijing time on 1 November 2026 (16:00 UTC on 31 October) and have since been billed at $0.75 per 1,000 successful requests (DeepSeek’s Search guide, checked 12 November 2026). Intel Macs are not covered by that build and should use the web chat. For an iOS-specific walkthrough of the chat app, see DeepSeek on iPhone.
Option 3: Run DeepSeek locally on Mac with Ollama or LM Studio
Local inference is where Apple Silicon earns its keep. The unified memory architecture on M-series chips means a 24 GB MacBook Pro can hold model weights that would need a discrete GPU on a PC. The catch: DeepSeek’s current flagship models are far larger than typical Mac memory configurations. For scale, the April 2026 V4 preview model card listed 1.6T total parameters for V4-Pro (Preview), and DeepSeek’s 10 September 2026 announcement described the September V4.1-Flash build as a 552B-parameter MoE. Local users run the older R1 distilled models or the original DeepSeek Coder family.
Recommended local models by Mac
| Mac configuration | Recommended model | Ollama tag | Download size (Q4_K_M) |
|---|---|---|---|
| M1/M2/M3 8 GB | R1-Distill-Qwen-1.5B | deepseek-r1:1.5b |
1.1 GB |
| M-series 16 GB | R1-Distill-Qwen-7B | deepseek-r1:7b |
4.7 GB |
| M-series 24–32 GB | R1-Distill-Llama-8B or R1-Distill-Qwen-14B | deepseek-r1:8b-llama-distill-q4_K_M / deepseek-r1:14b |
4.9 GB / 9.0 GB |
| M-Pro/M-Max 64 GB | R1-Distill-Qwen-32B | deepseek-r1:32b |
20 GB |
| M-Ultra / Studio 128 GB+ | R1-Distill-Llama-70B | deepseek-r1:70b |
43 GB |
Sizes are the download sizes Ollama lists for its deepseek-r1 tags (checked 23 October 2026); memory use while a model runs is higher, because the context needs room too. Note that the plain deepseek-r1:8b tag now points to DeepSeek-R1-0528-Qwen3-8B (5.2 GB), not the Llama-based distill.
For a sizing check before you download 43 GB of weights, run the numbers through the DeepSeek hardware calculator.
Ollama install on macOS
Ollama is the path of least resistance. Install it once, then pull models on demand. In Terminal:
brew install ollama
ollama serve &
ollama pull deepseek-r1:7b
ollama run deepseek-r1:7b
That gives you a local REPL. Models are stored in ~/.ollama/models. Ollama exposes a local OpenAI-compatible API on http://localhost:11434/v1, so any tool that speaks OpenAI’s wire format will work against it. The full walkthrough — including model variants and GPU offload settings — is in the running DeepSeek on Ollama tutorial.
LM Studio for a graphical option
LM Studio is a native macOS app with a model browser, chat UI, and a built-in local server. It is friendlier than Ollama if you prefer not to live in Terminal, and it surfaces quantisation options clearly. Same underlying GGUF weights, same local-only privacy posture. For a broader local setup that covers both, see install DeepSeek locally.
Realistic expectations: speed depends on the chip, its memory bandwidth, the model size and the quantisation, and larger distills respond noticeably more slowly than small ones. The distilled models are far smaller than DeepSeek’s current cloud models, so treat them as a tool for privacy-first workflows and offline coding assistance rather than a stand-in for deepseek-flash or deepseek-v4-pro.
Option 4: The DeepSeek API from your Mac

If you are building anything — a script, a CLI tool, a Raycast extension, a custom Obsidian plugin — the API is the right surface. DeepSeek’s pricing page lists two models: deepseek-flash, served by DeepSeek-V4.1-Flash-1017 and the only current model with vision (image input), and deepseek-v4-pro, served by DeepSeek-V4-Pro-0813. Both have a 1M-token context window and a maximum output of 384K tokens.
Chat requests hit POST /chat/completions, the OpenAI-compatible endpoint at https://api.deepseek.com. DeepSeek also offers an Anthropic-compatible format at https://api.deepseek.com/anthropic, so you can swap in the Anthropic SDK if your codebase already uses it.
Quickstart on macOS
Install the OpenAI Python SDK in a virtualenv, then point it at DeepSeek. Add the export line to ~/.zshrc if you want the key available in every new Terminal window:
python3 -m venv ~/.venvs/deepseek
source ~/.venvs/deepseek/bin/activate
pip install openai
export DEEPSEEK_API_KEY="sk-..."
A minimal Python script:
import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.deepseek.com",
api_key=os.environ.get("DEEPSEEK_API_KEY"),
)
resp = client.chat.completions.create(
model="deepseek-flash",
messages=[{"role": "user", "content": "Summarise the M4 Max in one paragraph."}],
reasoning_effort="high",
extra_body={"thinking": {"type": "enabled"}},
)
print(resp.choices[0].message.content)
Two things to remember. First, the API is stateless — DeepSeek does not store turns server-side. Your client must resend the full messages array each request. Second, thinking mode is a parameter on either model, not a separate ID, and it is on by default; the script above writes it out explicitly. The response returns reasoning_content alongside the final content. For more depth use reasoning_effort="max" (or "low" for simple tasks), and to switch thinking off pass extra_body={"thinking": {"type": "disabled"}}. The full DeepSeek API documentation covers the parameter set in detail.
Useful parameters and features
temperature— DeepSeek’s official guidance: 0.0 for code/maths, 1.0 for data analysis, 1.3 for general conversation and translation, 1.5 for creative writing. It only takes effect in non-thinking mode.max_tokens— cap on output length; up to 384K on both models. Always set this when using JSON mode.response_format={"type": "json_object"}— JSON mode is designed to return valid JSON, not guaranteed. Include the word “json” and a small example schema in your prompt; handle occasional empty content.- Tool calling, streaming (
stream=true), context caching, image input ondeepseek-flash, FIM completion (Beta — requiresthinking: {"type": "disabled"}), and Chat Prefix Completion (Beta) are all supported.
Pricing for API use from a Mac
API costs depend on the model and on when you send requests. As of October 2026, per DeepSeek’s official pricing page, deepseek-flash costs $0.006 (cache hit) / $0.30 (cache miss) per 1M input tokens and $1.20 per 1M output tokens at peak, and $0.003 / $0.15 / $0.60 off-peak. deepseek-v4-pro costs $0.044 / $1.32 / $3.96 at peak and $0.022 / $0.66 / $1.98 off-peak. Off-peak rates are half of peak; peak hours are 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday, excluding Chinese public holidays. Worked examples are on our DeepSeek API pricing page — always confirm current rates there before committing.
Common Mac-specific issues and fixes
| Symptom | Likely cause | Fix |
|---|---|---|
| Ollama stalls on first prompt | Model still loading into memory | Wait 10–30 seconds; check Activity Monitor for memory pressure |
| “Out of memory” on M1 8 GB | Model too large for unified RAM | Drop to a smaller distill (1.5B) or use the API instead |
| DeepSeek chat app not found in the Mac App Store | Its listing shows iPhone, iPad and iPod touch compatibility only | Use the web chat in Safari; on Apple silicon, DeepSeek Harness has a macOS build |
| API request returns 401 | Missing or wrong key | Re-export DEEPSEEK_API_KEY; rotate in the console |
| Empty JSON response | Model truncation or prompt without “json” | Raise max_tokens; add example schema to prompt |
For a broader catalogue, see the DeepSeek troubleshooting guide.
Privacy and where your data goes

The web, app and API all send conversations to DeepSeek’s infrastructure in China. That is fine for casual use; it is not appropriate for client-confidential legal work, medical records, or anything covered by GDPR data-residency clauses unless your employer has explicitly approved it. The local Ollama or LM Studio path keeps every byte on your Mac — that’s the privacy trade-off worth the extra setup time. The DeepSeek privacy guide goes into the contractual details.
Which path should you pick?
- Casual user, just curious: the web chat. Move on with your day.
- Mac power user wanting offline access: Ollama with R1-Distill-Qwen-7B on 16 GB or larger.
- Developer building tools:
deepseek-flashvia the API; trydeepseek-v4-proon specific calls where you want to compare results. - Privacy-sensitive work: local only; do not use the cloud surfaces.
For a more general beginner pathway, the DeepSeek beginners guide is a softer entry point. If you want to compare DeepSeek’s Mac story against ChatGPT’s, the DeepSeek vs ChatGPT comparison covers ecosystem differences honestly. And for a curated index of every guide on this site, the DeepSeek beginner guides hub is the catalog.
Is there a native DeepSeek app for macOS?
For chat, no: as of 23 October 2026 DeepSeek’s download page offers the chat app for iOS and Android, and its App Store listing gives iPhone, iPad and iPod touch compatibility without listing Mac, so use the web chat. DeepSeek does offer a desktop app of a different kind — DeepSeek Harness, its agent app in public preview — with a macOS build for Apple silicon (macOS 13 or later). The DeepSeek app guide covers verification.
How do you run DeepSeek locally on a MacBook?
Install Ollama with brew install ollama, pull a distill model with ollama pull deepseek-r1:7b, then run it. A 16 GB Apple Silicon Mac handles 7B models comfortably; 32B needs 64 GB. DeepSeek’s current flagship models are far larger than typical Mac memory. Step-by-step setup lives in the install DeepSeek locally tutorial.
Can you use the DeepSeek API from a Mac terminal?
Yes. Install the OpenAI Python SDK with pip, set DEEPSEEK_API_KEY as an environment variable, and point base_url at https://api.deepseek.com with model="deepseek-flash" or "deepseek-v4-pro". Chat requests hit POST /chat/completions. The API is stateless, so resend message history each call. The DeepSeek API getting started tutorial walks through the full setup.
What hardware does DeepSeek on Mac need?
For the web chat, any modern Mac works; DeepSeek Harness needs Apple silicon and macOS 13 or later. For local inference, Apple Silicon with 16 GB unified memory handles 7B-class distill models; 32 GB handles 14B; 64 GB handles 32B comfortably. Intel Macs can run small distill models on CPU but performance is poor. Check sizing in the DeepSeek hardware calculator before downloading weights.
Does DeepSeek on Mac cost anything?
The web chat and the official chat app are free, and DeepSeek does not document a daily message cap. Local Ollama and LM Studio are free aside from electricity. The API charges per token: as of October 2026, deepseek-flash is $0.30 input (cache miss) / $1.20 output per million tokens at peak and $0.15 / $0.60 off-peak; deepseek-v4-pro is $1.32 / $3.96 at peak and $0.66 / $1.98 off-peak. Always confirm on the DeepSeek API pricing page before budgeting.
Sources & Methodology
Primary sources
- RepositoryDeepSeek-AI on GitHubOpen-weight release details, training/inference notesLast checked: November 12, 2026
- OfficialDeepSeek — Download pageDeepSeek Harness desktop for macOS (Apple silicon, macOS 13+), chat app for iOS/Android, free chat; accessed 23 Oct 2026Last checked: November 12, 2026
- PricingDeepSeek API Docs — Models & PricingModels, vision on deepseek-flash, 1M context / 384K output, peak and off-peak prices, peak hours, concurrency, Harness as an API surface; accessed 23 Oct 2026Last checked: November 12, 2026
- OfficialDeepSeek API Docs — Your First API CallBase URLs and thinking fields in the sample code; accessed 23 Oct 2026Last checked: November 12, 2026
Technical references
- Technical reportDeepSeek-R1 paper (arXiv:2501.12948)R1 reasoning approach, distillation resultsLast checked: November 12, 2026
- OfficialDeepSeek API Docs — Thinking ModeThinking default, reasoning_effort, temperature ignored in thinking mode; accessed 23 Oct 2026Last checked: November 12, 2026
- App store listingApp Store — DeepSeek - AI AssistantSeller name; compatibility lists iPhone, iPad, iPod touch (no Mac); accessed 23 Oct 2026Last checked: November 12, 2026
- OfficialDeepSeek Harness — QuickstartHarness uses a DeepSeek API key entered under Settings → Models; other providers supported; accessed 23 Oct 2026Last checked: November 12, 2026
- Vendor documentationOllama — deepseek-r1 tagsR1-distill tag names and Q4_K_M download sizes; accessed 23 Oct 2026Last checked: November 12, 2026
Context sources
- OfficialDeepSeek API Docs — Change Log17 Oct 2026 web/app model update to DeepSeek-V4.1-Flash-1017; accessed 23 Oct 2026Last checked: November 12, 2026
- OfficialDeepSeek News — DeepSeek-V4.1-Flash release (10 Sep 2026)552B-parameter MoE description of V4.1-Flash (dated, attributed); accessed 23 Oct 2026Last checked: November 12, 2026
- OfficialDeepSeek-V4-Pro-0813 model cardBuilt on the V4-Pro (Preview) structure; MIT weights; accessed 23 Oct 2026Last checked: November 12, 2026
- OfficialDeepSeek homepageDeepThink and Search options shown; accessed 23 Oct 2026Last checked: November 12, 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.
