How do I use DeepSeek in Codex CLI?
Run standardcompute launch codex -- --model deepseek/deepseek-v4.1-flash, or add a Responses provider with base_url https://api.stdcmpt.com/v1 to ~/.codex/config.toml: Codex then runs on DeepSeek V4.1 Flash on a plan from $19 a month, and new requests get HTTP 402 when the plan's compute budget reaches 100%.
# Install Codex and the Standard Compute launcher once
npm install -g @openai/codex
curl -fsSL https://standardcompute.com/cli/install.sh | sh
# Start Codex pinned to deepseek/deepseek-v4.1-flash
standardcompute launch codex -- --model deepseek/deepseek-v4.1-flash
The launcher leaves your own Codex config alone and lists every served model under /model. The DeepSeek tab follows DeepSeek's Codex guide.
What works with DeepSeek in Codex
| Codex feature | On DeepSeek through Standard Compute |
|---|---|
| Shell commands and file edits | Codex's tools reach DeepSeek through the Responses API |
| Images | Work on deepseek/deepseek-v4.1-flash; deepseek/deepseek-v4-pro is text-only and answers an image with HTTP 400 |
| Web search | Off: the launcher disables Codex's built-in web search on this connection |
| /model | With the launcher, lists every served model, so you can switch to V4 Pro or a GPT model mid-session |
| Reasoning | The launcher offers one level, medium, for every model |
| Your ChatGPT plan | Not used; requests bill to the Standard Compute plan |
From the launcher source and Standard Compute's Codex docs, October 14, 2026.
What a day of DeepSeek costs in Codex
| Per day | DeepSeek API, peak | Off-peak |
|---|---|---|
| Typical day | $0.42 | $0.21 |
| Busy day | $4.31 | $2.16 |
DeepSeek list prices per million tokens for V4.1 Flash: $0.30 input, $0.006 cache hit, $1.20 output at peak, half off-peak. Source. A typical day is 8.5M tokens and a busy one 86.7M, 87.2% of input as cache reads: the median and 90th-percentile coding-agent day in our Agent Burn Index, measured on OpenCode, Cline, Kilo and similar agents, September 13 to October 12, 2026. Priced at the list rates above; your own days will differ.
The catch
DeepSeek documents its own Codex setup: a Responses provider at https://api.deepseek.com/ with deepseek-flash or deepseek-v4-pro. If DeepSeek is the only model you want in Codex, that route costs less than any plan. A Standard Compute key earns its place when Codex should switch between DeepSeek, GPT and Claude models without separate accounts.
Where Standard Compute fits
The launcher sets everything for one session and leaves your own Codex config alone. At peak list price the Starter ($19 a month, $20 of compute) plan covers about 47 typical V4.1 Flash days, and spending stops there.
FAQ
Does Codex still support wire_api = "chat" for DeepSeek?
No. Current Codex releases talk to custom providers only through the Responses API. Both DeepSeek's endpoint and Standard Compute serve it, so set wire_api = "responses".
Why does Codex return 404 with my DeepSeek config?
Usually the base URL or API flavor. On Standard Compute, base_url must end in /v1 and wire_api must be "responses".
Can I use DeepSeek V4 Pro in Codex?
Yes: deepseek/deepseek-v4-pro. It is text-only and costs about 5 times as much per day as V4.1 Flash at peak, so switch to it with /model for hard steps.