One-Time DeepSeek Access

DeepSeek Coding Plan

Use the new native multimodal deepseek-flash model for coding, screenshots, charts, and mixed text-image tasks, then copy the correct setup for your coding agent.

Available while stock lasts

One access Key for V4.1 Flash and the existing V4 routes

Pay once for 30 days of API access. New configurations should use deepseek-flash for both text and image input. Existing configurations continue to work with the same encrypted Key and Base URL while you migrate. Ready-to-copy Agent configurations appear in Dashboard.

$39.90 · one-time

Review Agent setup

Delivery

One access key for the new and compatibility routes

5-hour limit

50M Flash-equivalent tokens

Long-window limit

400M tokens / 7 days

Access term

30 days · one-time

1 · How to choose

Use deepseek-flash by default for both text and images

The new V4.1 Flash route has a 1M context window, supports native multimodal input, and produces up to 384K output tokens. All routes share one token pool. Flash routes use one plan token per raw input/output token reported by the provider. The Pro compatibility route uses ceil(tokens × 87 / 28) plan tokens, or about 3.11×, so a 50M five-hour pool represents about 16.09M raw Pro tokens if every request uses Pro. Send images by URL or Base64; this plan does not include Files API uploads.

DeepSeek V4.1 Flash

Default · native vision · 1×

deepseek-flash

Use the new native multimodal model for everyday coding, screenshots, charts, and mixed text-image Agent work. Send image URLs or Base64 input through the supported API protocol.

DeepSeek V4 Flash

Compatibility · 1×

deepseek-v4-flash

Keep this existing text route for compatibility with configurations that have not yet moved to deepseek-flash.

DeepSeek V4 Pro

Compatibility · hard tasks · ≈3.11×

deepseek-v4-pro

Switch to Pro for difficult reasoning, architecture decisions, deep reviews, and final-pass verification where quality matters more than token volume.

DeepSeek V4 Flash Vision Exp

Compatibility · image input · 1×

deepseek-v4-flash-vision-exp

Keep this experimental Vision route for compatibility with existing image configurations. New configurations should use the native multimodal deepseek-flash model.

2 · Switch models

Keep the Key and Base URL; change only model

Start with deepseek-flash. For a new image task, keep the same model and add a URL or Base64 image to the request. The older deepseek-v4-flash, deepseek-v4-pro, and deepseek-v4-flash-vision-exp IDs remain compatibility routes, so existing configurations continue to work while users move to the new ID.

// New default
model: "deepseek-flash"

// Native image input
model: "deepseek-flash"

// Existing configs
model: "deepseek-v4-flash"

3–4 · Protocol, test, then Agent

The Key and models are shared; the wire contracts are not

Select the exact protocol your client sends. Each option below has a different endpoint, request body, and response shape, followed by a matching API test and Agent configuration.

These examples start with deepseek-flash and already use the API endpoint provided with the Token Plan athttps://api.deepseekv4pro.com/v1. Replace only YOUR-TOKEN-PLAN-KEY with the access Key delivered in Dashboard. This product does not transfer a DeepSeek account or credentials from the DeepSeek console.

Protocol A · Responses

OpenAI Responses

Public route configured

Base URL: https://api.deepseekv4pro.com/v1
Exact endpoint: https://api.deepseekv4pro.com/v1/responses
Authentication: Authorization: Bearer <access key>

Request → response contract

input request → output items and Responses events

Compatible tools

Codex

Do not mix /v1/responses, /v1/chat/completions, and /v1/messages. The Key and model aliases are shared; the wire contracts are not.

API connection test

Expected marker: responses-ok

curl https://api.deepseekv4pro.com/v1/responses \
  -H "Authorization: Bearer YOUR-TOKEN-PLAN-KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-flash",
    "input": [
      {"role": "developer", "content": [{"type": "input_text", "text": "Answer concisely."}]},
      {"role": "user", "content": [
        {"type": "input_text", "text": "Describe this image, then reply with: responses-ok"},
        {"type": "input_image", "image_url": "https://example.com/image.png"}
      ]}
    ],
    "max_output_tokens": 32
  }'

Agent configuration for this protocol

Codex

~/.codex/config.toml

Open configuration

Install

npm install -g @openai/codex

Launch

codex

OpenAI Responses only

model = "deepseek-flash"
model_provider = "deepseek-token-plan"

[model_providers.deepseek-token-plan]
name = "DeepSeek Coding Plan"
base_url = "https://api.deepseekv4pro.com/v1"
env_key = "DEEPSEEK_TOKEN_PLAN_KEY"
wire_api = "responses"

# Put this in the shell that launches Codex:
# export DEEPSEEK_TOKEN_PLAN_KEY="YOUR-TOKEN-PLAN-KEY"

Switch model

Use deepseek-flash for new text or image tasks. Change only the top-level model value to a legacy V4 ID when an existing workflow still depends on that compatibility route, then start a new Codex task.

Verify

Run codex with a small read-only prompt and confirm the request reaches the Responses route without an authentication or model error.

Requires an OpenAI Responses-compatible API route. The provided endpoint exposes /v1/responses for that purpose.

Official client documentation

Purchase strategy

From model choice to first Agent request

01 · Choose

Start with deepseek-flash for native text and image tasks; keep a legacy route only when an existing workflow requires it.

02 · Buy

Pay $39.90 once through Stripe for a 30-day access term while inventory remains.

03 · Receive

The encrypted access key and Base URL are assigned in Dashboard after payment.

04 · Connect

Choose the model and Agent tab, copy the generated config, then run the verification prompt.