Recommended Free Tools
You can keep using Claude Code or Codex as your coding-agent interface and route model requests to DeepSeek. The setup differs: Claude Code uses DeepSeek’s Anthropic-compatible endpoint, while Codex uses its Responses API configuration. Both routes require a DeepSeek API key.
Use DeepSeek’s current setup guides for exact commands and supported client versions; provider settings and model identifiers can change.
What changes—and what stays the same
Claude Code or Codex remains the client you use to work in your project. The change is the model provider those clients contact. This is not one shared setup: Claude Code and Codex use different API routes and configuration methods.
| Client | DeepSeek API route | Configuration approach |
|---|---|---|
| Claude Code | Anthropic-compatible endpoint: https://api.deepseek.com/anthropic |
Set environment variables for the endpoint, API token, and model slots, then launch claude from your project directory. See DeepSeek’s Claude Code integration guide. |
| Codex | Responses API, using https://api.deepseek.com/ |
Use DeepSeek’s setup script or merge provider settings into Codex configuration. See the Codex integration guide and DeepSeek’s Responses API documentation. |
DeepSeek describes its API format as compatible with OpenAI and Anthropic, but that does not mean every client uses the same protocol for its integration. Follow the route documented for your client.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →#1 Best Overall
Before you configure either client
- Get a DeepSeek API key. The documented configurations use the key to authenticate API requests; a ChatGPT or Claude subscription is not a substitute for it.
- Check DeepSeek’s current guide for the supported client version and model identifiers. Names and routing can change.
- Keep your API key private. Treat it as a credential, not as a value to commit to a project repository or share in a prompt.
Set up DeepSeek in Claude Code
DeepSeek’s Claude Code instructions configure Claude Code to use its Anthropic-compatible endpoint. The guide provides environment-variable examples for Linux/macOS and Windows, including the endpoint, authentication token, and model selections.
- Open DeepSeek’s Claude Code integration guide and choose the instructions for your operating system.
- Set
ANTHROPIC_BASE_URLtohttps://api.deepseek.com/anthropicand provide your DeepSeek API key throughANTHROPIC_AUTH_TOKEN, following the guide’s exact syntax for your shell. - Configure the model variables shown in the live guide. Its documented examples include
ANTHROPIC_MODELand the default Opus, Sonnet, and Haiku slots; the listed values includedeepseek-flash[1m]for the first three anddeepseek-flashfor Haiku and subagents. - Enter your project directory and run
claude.
The guide also sets CLAUDE_CODE_EFFORT_LEVEL=max. Use the live instructions rather than copying an older snippet: the current page does not show the ANTHROPIC_CODE variable found in some older versions of the setup.
Set up DeepSeek in Codex
DeepSeek documents a Codex integration using the Responses API. Its guide offers a one-click setup script for macOS/Linux and PowerShell, as well as a manual configuration path.
Use the setup script
Follow the platform-specific script instructions in DeepSeek’s Codex guide. The guide says the script backs up existing configuration, validates files before writing, creates or updates ~/.codex/models.json and ~/.codex/config.toml, and provides a restore option. Review the current script and its prompts before running it, especially if your Codex settings are customized.
Rank #3
Configure Codex manually
DeepSeek’s example selects deepseek-flash, sets wire_api = "responses", and uses the base URL https://api.deepseek.com/. Follow the full current example in the official guide for the required provider and model fields. If you already have a config.toml, merge the settings into existing sections rather than adding duplicate TOML sections.
The guide says Codex CLI, the ChatGPT desktop app, and the VS Code Codex extension share this configuration. It also says the script preserves MCP and project-trust settings; check the current guide and script behavior rather than assuming this will always be true.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Choose a current DeepSeek model name
DeepSeek’s API overview currently lists deepseek-flash and deepseek-v4-pro. It says the legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp remain accepted but are retired aliases routed to DeepSeek-V4.1-Flash and billed at the Flash price. Prefer the current canonical names in new configuration, and confirm the live model list before changing a working setup. See DeepSeek’s API guide.
DeepSeek’s Codex guide lists a context-window setting of 1,048,576 tokens for its example model configuration. That is model-catalog metadata in the guide accessed in 2026, not a performance measurement; check the current guide before relying on it for a model.
Best Value
If Codex sessions seem to disappear
DeepSeek’s Codex guide says session histories are grouped by login method. After switching provider configuration, Codex may display only the history group associated with the active login. The guide does not say that the other group is deleted; restoring the previous configuration makes the earlier group visible again.
What this setup does not establish
The official integration instructions explain how to configure the clients; they are not independent tests of coding quality or reliability. They also do not establish current API pricing, data-retention practices, regional availability, or usage limits. Check DeepSeek’s current official information for those questions rather than inferring answers from the setup guides.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
