Claude Code
DeepSeek documentedThe mature terminal route for DeepSeek-backed coding, explicit Pro and Flash routing, web search, and subagent work.
DeepSeek Guides by Topic
Start with the job you need to complete: choose a coding agent, build against the API, use reasoning features correctly, operate or run a model locally, or compare cost and model fit.
Coding agents
Agents & Coding Tools
Choose and configure Claude Code, DeepSeek TUI, Reasonix, Codex-compatible routes, and other DeepSeek-backed agent tools.
DeepSeek TUI
Independent projectThe CodeWhale terminal interface for users who want DeepSeek-focused modes, configuration, and project controls.
Reasonix
DeepSeek documentedA DeepSeek-native coding loop built around Flash-first iteration, cache-aware sessions, and explicit Pro escalation.
Codex
Compatibility watchAn OpenAI custom-provider route for teams already using Codex and prepared to verify DeepSeek protocol and model compatibility.
Other Agent Guides (18)
OpenCode, OpenClaw, Deep Code, GitHub Copilot, Hermes, nanobot, Kilo Code, Pi, Crush, AstrBot, Langcli, and related support guides remain available without competing with the four primary routes above.
Set Up DeepSeek in GitHub Copilot ChatSet Up GitHub Copilot CLI With DeepSeekUse DeepSeek Copilot Chat With a Vision ProxyFix the DeepSeek Copilot CLI reasoning_content 400 ErrorSet Up Hermes Agent With DeepSeek V4 ProSet Up Deep Code CLI With DeepSeekUse Project and Global Skills in Deep CodeSet Up nanobot With DeepSeek V4 ProSet Up OpenCode With DeepSeek V4 ProConfigure Crush for DeepSeek V4 Pro and FlashConfigure Pi for DeepSeek V4 ReasoningConfigure WorkBuddy and CodeBuddy for DeepSeek V4Set Up Kilo Code With DeepSeek V4 Pro or FlashConfigure Oh My Pi for DeepSeek Thinking ModeSet Up OpenClaw With DeepSeek V4 Pro or FlashConnect DeepSeek OpenClaw to WeChat and FeishuSet Up AstrBot With DeepSeekDeepSeek Langcli Guide: official install, LangRouter flow, and routing tradeoffs
API & SDK
Build against DeepSeek with Python, Node.js, OpenAI- or Anthropic-compatible APIs, request history, model discovery, and balance checks.
Updated 2026-05-24
How to Use DeepSeek in Python — From Zero to Production
Step-by-step guide to calling DeepSeek V4 from Python: install the OpenAI SDK, configure the base URL, stream responses, use function calling, manage tokens, and cut costs with discounted official API keys.
Read guideUpdated 2026-05-24
How to Use DeepSeek in Node.js — TypeScript, Streaming, Tool Use
Production-ready tutorial for calling DeepSeek V4 from Node.js and TypeScript using the OpenAI SDK: setup, streaming, function calling, Next.js server-side usage, and cost control tips.
Read guideUpdated 2026-06-10
DeepSeek OpenAI vs Anthropic API Routing
DeepSeek-first routing guide for choosing between the OpenAI-format and Anthropic-format DeepSeek endpoints, with model-name rules, migration deadlines, request examples, and verification checklists for coding agents.
Read guideUpdated 2026-06-27
Replay Full History With DeepSeek Chat Completions
Official DeepSeek guide to multi-round chat history: `/chat/completions` is stateless, every follow-up call must resend prior turns, and the assistant response from the last round belongs in `messages` before the next user question.
Read guideUpdated 2026-07-01
Create a DeepSeek Chat Completion
Official DeepSeek Create Chat Completion reference for API users: required `messages`, supported message roles, streaming SSE chunks, `finish_reason` meanings, and how to read usage and tool-call responses safely.
Read guideUpdated 2026-07-02
How to use DeepSeek's List Models API without guessing which model IDs are live
Official DeepSeek guide to the `GET /models` endpoint: what the response contains, how it differs from the pricing page, when to use it for runtime validation, and how to keep V4 model routing clean before old aliases disappear.
Read guideUpdated 2026-07-02
How to read DeepSeek's User Balance API before your workflows run out of budget
Official DeepSeek guide to `GET /user/balance`: what `is_available` means, how `granted_balance` differs from `topped_up_balance`, which currencies the docs list, and how to use the endpoint for budget checks without confusing it with token pricing.
Read guideReasoning & Model Features
Use thinking mode, tool calls, reasoning content, context caching, structured output, FIM, and reusable prompt patterns correctly.
Updated 2026-06-11
DeepSeek Thinking Mode & Tool Calls in Multi-Turn Apps
Official DeepSeek guide to thinking mode, reasoning effort, `reasoning_content`, tool-call replay rules, and multi-turn handling for OpenAI-format and Anthropic-format V4 apps.
Read guideUpdated 2026-06-25
DeepSeek Thinking Mode Parameters: stop tuning knobs that the model ignores
Official DeepSeek guide to thinking-mode parameter rules: unsupported sampling knobs, OpenAI vs Anthropic effort controls, `reasoning_content` replay, and practical debugging for agent stacks.
Read guideUpdated 2026-06-26
DeepSeek Context Caching and Prefix Reuse
Official DeepSeek context caching guide for API users: when prefixes persist, why second requests can still miss, how `prompt_cache_hit_tokens` works, and how to structure repeated prompts so DeepSeek V4 Pro or Flash actually gets cheaper.
Read guideUpdated 2026-06-27
DeepSeek JSON Output guide: use the official contract, not wishful prompting
Official DeepSeek JSON Output guide for structured responses: set `response_format` to `json_object`, say "json" in the prompt, give an example schema, cap `max_tokens` sensibly, and watch for occasional empty-content returns.
Read guideUpdated 2026-06-29
Use DeepSeek Chat Prefix Completion (Beta)
Official DeepSeek Chat Prefix Completion beta guide: the last message must be an assistant message with `prefix: true`, the Beta route requires `https://api.deepseek.com/beta`, and `stop` helps keep code-only completions clean.
Read guideUpdated 2026-06-29
Use DeepSeek FIM Completion With Prefix and Suffix
Official DeepSeek FIM Completion beta guide: send a prefix and optional suffix to fill in the middle, use `https://api.deepseek.com/beta`, respect the 4K max-token cap, and understand where Continue fits.
Read guideUpdated 2026-06-30
DeepSeek Temperature Parameter Guide: use the official range for the actual task
Official DeepSeek temperature guide for API users: default `temperature` is 1.0, coding and math are recommended at 0.0, data cleaning at 1.0, conversation and translation at 1.3, and creative writing at 1.5.
Read guideUpdated 2026-07-01
DeepSeek Reasoning Model Inputs and Parameter Limits
Official DeepSeek reasoning-model guide for API users: `max_tokens` counts the CoT, unsupported sampling parameters are ignored or rejected, and replaying `reasoning_content` back into input messages returns a 400 error.
Read guideUpdated 2026-07-06
DeepSeek Tool-Call Python Example
Use DeepSeek's official Python tool-call sample to build function calling with thinking mode, structured tool schemas, multi-step execution, and safe replay rules.
Read guideUpdated 2026-07-06
DeepSeek reasoning_content Output and Replay
Study DeepSeek's official reasoning_content output transcript to understand how thinking mode, tool calls, and final answers change across turns without breaking replay logic.
Read guideUpdated 2026-07-08
DeepSeek Non-Streaming Thinking Mode Sample
Read DeepSeek's official non-streaming thinking-mode sample to understand the two-turn request pattern, how `reasoning_content` is returned, and why the API ignores it in ordinary follow-up turns.
Read guideUpdated 2026-07-08
DeepSeek Streaming Thinking Mode Sample
Use DeepSeek's official streaming thinking-mode sample to understand how `delta.reasoning_content` and `delta.content` arrive, how to accumulate them safely, and how the second turn is structured.
Read guideUpdated 2026-07-13
Use the Official DeepSeek Prompt Library
DeepSeek-first guide to the official Prompt Library: what the first-party page actually is, how to mine reusable patterns, how to adapt examples safely, and how to connect prompt experiments back to real V4 workflows.
Read guideOperations & Local Deployment
Migrate model names, diagnose API failures, manage concurrency, verify official updates, and run DeepSeek V4 Flash or GGUF builds locally.
Updated 2026-06-09
DeepSeek Chat Deprecation Guide: Move to V4 Before July 24, 2026
Official DeepSeek migration guide for replacing deepseek-chat and deepseek-reasoner with DeepSeek V4 Pro or Flash across OpenAI SDK code, Claude Code, and production rollout checklists.
Read guideUpdated 2026-06-11
DeepSeek Rate Limit and user_id Isolation: production rules for V4 traffic
Official DeepSeek guide to concurrency limits, `user_id` isolation, 429 handling, keep-alive behavior, and production routing for multi-user V4 deployments.
Read guideUpdated 2026-06-28
DeepSeek V4 Concurrency Limits for Flash and Pro
Understand DeepSeek's official V4 concurrency contract: account-level limits of 2500 for Flash and 500 for Pro, what triggers HTTP 429, when `user_id` matters, and how to request higher capacity safely.
Read guideUpdated 2026-06-30
DeepSeek API Error Codes and Fixes
Official DeepSeek error-code guide for API users: what 400, 401, 402, 422, 429, 500, and 503 mean, which failures are request bugs versus account issues versus provider load, and how to recover without guessing.
Read guideUpdated 2026-06-26
Migrate DeepSeek Chat and Reasoner Before July 24, 2026
Official DeepSeek migration guide for `deepseek-chat` and `deepseek-reasoner`: what retires on July 24, 2026, which aliases map to Flash today, when to move to V4 Pro, and how to update OpenAI and Anthropic-compatible clients safely.
Read guideUpdated 2026-07-05
Fix DeepSeek Chat and Reasoner 404 Errors
Official DeepSeek troubleshooting guide for `deepseek-chat` and `deepseek-reasoner` failures: why the aliases are being retired, which V4 model names replace them, how to validate `/models`, and how to stop old configs from breaking production.
Read guideUpdated 2026-07-05
Replace deepseek-reasoner With V4 Flash Thinking Mode
Official DeepSeek migration guide for teams replacing `deepseek-reasoner`: how V4 Flash thinking mode maps to the legacy alias, when Pro is still worth it, and how to update OpenAI-compatible and Anthropic-compatible clients safely.
Read guideUpdated 2026-07-14
DeepSeek V4 Flash GGUF: Mac Setup Guide
Run DeepSeek V4 Flash GGUF on a Mac: check memory, choose a quantization, build a compatible llama.cpp runtime, launch it, and validate the output.
Read guideUpdated 2026-07-14
DeepSeek V4 Pro GGUF: Is There an Official Download?
DeepSeek publishes V4 Pro weights, but not an official GGUF file. See which files are official, how community quantizations differ, and what can run locally.
Read guideUpdated 2026-06-15
How to Verify DeepSeek Official X Updates
A DeepSeek-first workflow for tracking official X/Twitter updates when X is login-blocked: verify the homepage anchor, then confirm with DeepSeek docs, status, GitHub, and Hugging Face before publishing or sharing claims.
Read guideComparisons, Pricing & Plans
Compare DeepSeek models and competitors, understand API pricing, and keep Coding Plan content aligned with real inventory.
Updated 2026-06-09
DeepSeek V4 Pro vs Flash for Coding, Agents, and Claude Code
DeepSeek-first guide to choosing DeepSeek V4 Pro or Flash for coding, agents, Claude Code, and API routing using official pricing, context, and model documentation.
Read guideUpdated 2026-05-08
DeepSeek V4 vs MiniMax 2.7
Detailed DeepSeek V4 vs MiniMax 2.7 comparison for developers: coding quality, agent workflows, API compatibility, local deployment options, context strategy, pricing risk, and when to use each model.
Read guideUpdated 2026-05-24
DeepSeek V4 vs Claude (Sonnet 4 / Opus 4)
In-depth comparison of DeepSeek V4 vs Anthropic Claude Sonnet 4 and Opus 4: coding, reasoning, long-context, agentic tool use, safety, and price. Includes when to use each and how to combine them.
Read guideUpdated 2026-05-24
DeepSeek V4 vs ChatGPT (GPT 5.4)
DeepSeek V4 vs OpenAI ChatGPT (GPT 5.4): coding quality, reasoning, context window, price, ecosystem, and when each model wins. Includes migration tips for teams already on OpenAI.
Read guideUpdated 2026-05-24
DeepSeek V4 vs Grok (Grok 4 / Grok Heavy)
DeepSeek V4 vs xAI Grok 4 and Grok Heavy: coding, reasoning, real-time knowledge, context window, pricing, and which one to pick for product teams versus power users.
Read guideUpdated 2026-06-28
DeepSeek V4 Pricing per Million Tokens
Use DeepSeek's official pricing contract to compare DeepSeek V4 Flash and Pro per 1M tokens, understand cache hit versus cache miss billing, and estimate when Flash or Pro makes sense for real workloads.
Read guideUpdated 2026-05-08
DeepSeek V4 Pro Coding Plan: Availability and Safer Buying Rules
SEO guide for DeepSeek V4 Pro Coding Plan searchers. Explains what a Pro Coding Plan would mean, why it is not listed as a purchasable product unless inventory is secured, and how to choose current in-stock alternatives.
Read guide