The DeepSeek coding plan with no token caps.
DeepSeek's own API bills per token. Kourier runs DeepSeek V4.1 Flash on our own GPUs and sells it as a flat monthly plan for coding agents: Starter at $20/month, Pro at $50/month.
Starter
$20/month
- No request-volume or token-usage quotas
- Up to 1 concurrent request
- DeepSeek V4.1 Flash access; new models added as available
- OpenAI-compatible API endpoint
- Community support via Discord
- Usage dashboard and request logs
Recommended
Pro
$50/month
- No request-volume or token-usage quotas
- Up to 3 concurrent requests
- DeepSeek V4.1 Flash access; new models added as available
- OpenAI-compatible API endpoint
- Community support via Discord
- Usage dashboard and request logs
Flat pricing, not a token meter.
Agent sessions are long and bursty. Kourier limits how many requests run at once, never how many tokens you use.
| Feature | Per-token API | Token-quota plan | Kourier |
|---|---|---|---|
| How you pay | Per million input and output tokens | Monthly fee for a token allowance | $20 or $50 per month, flat |
| What limits you | Your budget | Token windows, often every 5 hours and every week | Concurrent requests only: 1 on Starter, 3 on Pro |
| Long agent sessions | Every retry and tool call adds to the bill | Can stop mid-session when a window runs out | No 5-hour, weekly, or monthly cap |
| Monthly cost | Varies with usage | Fixed, plus top-ups when you run out | Fixed |
Works with your coding agent.
One OpenAI-compatible and Anthropic-compatible endpoint. Set a base URL, an API key, and the model deepseek-v4.1-flash.
DeepSeek coding plan questions.
DeepSeek's own API bills per token. Kourier is a separate service: we run DeepSeek V4.1 Flash on our own GPUs and sell it as a flat monthly plan built for coding agents.
DeepSeek V4.1 Flash, API model string deepseek-v4.1-flash. The maximum context length is 262,144 tokens (256K) for prompt plus output combined, with up to 16,384 tokens per response.
Concurrency. Starter accepts 1 simultaneous request and Pro accepts up to 3. When a request finishes, that slot is immediately available again; neither plan has a 5-hour, weekly, or monthly request or token cap.
Any tool that accepts a custom OpenAI-compatible or Anthropic-compatible endpoint, including Oh My Pi, Claude Code, Codex CLI, Cursor, Cline, OpenCode, Zed, Continue, and Aider. The docs have setup steps for each.
No. Sign up, choose Starter or Pro in Billing, and create an API key.
We do not use your API prompts or outputs to train or fine-tune models. Prompt and response content is excluded from inference gateway request logs. Requests are processed in memory with temporary caching. We retain operational metadata for usage, billing, security, and service reliability. Read our Data use & retention policy.
Start coding on DeepSeek.
No invite needed. Sign up, pick Starter or Pro, and create an API key.