DeepSeek-V4-Flash, 4× cheaper
The same open-weight DeepSeek-V4-Flash — fixed USD per-token, a 380K context window and room for long replies. Flash-class speed for high-volume agent and coding traffic.
Why run it here
Flat published USD rates per million tokens, about 4× below OpenRouter on the same model. Pay-as-you-go from a USD balance — a fast, cheap default for agents that burn through tokens all day.
Drop-in, OpenAI-compatible
Point any OpenAI-compatible client at our base URL and pick the DeepSeek-V4-Flash model id. Requests, streaming and tool calls pass straight through, with the whole 380K window available.
Use DeepSeek-V4-Flash
# DeepSeek-V4-Flash via cheaptokens (OpenAI-compatible)
export OPENAI_BASE_URL="https://api.cheaptokens.dev/v1"
export OPENAI_API_KEY="YOUR_CHEAPTOKENS_KEY"
# model: deepseek-ai/DeepSeek-V4-Flash-0731DeepSeek-V4-Flash vs OpenRouter
Per-million-token rates on the exact same model.
| Model | OpenRouter | cheaptokens | You save |
|---|---|---|---|
| DeepSeek-V4-Flash temporarily unavailable | $0.016 / $0.64 | $0.0221 / $0.10 | ~4× cheaper |
| GLM-5.3-Flash temporarily unavailable | $0.048 / $0.20 | $0.027 / $0.08 | ~2× cheaper |
Same models. USD per 1M tokens (input / output). OpenRouter shown for comparison; savings by blended price (3 parts input to 1 part output). · Open cost calculator
FAQ
Is this the same DeepSeek-V4-Flash?
Yes — the same open-weight DeepSeek-V4-Flash (0731 build), served at a fixed USD per-token rate.
How big is the context window?
380K tokens. Enough for whole repositories, long agent sessions and big document batches.
How do I pay?
Top up a USD balance with crypto — no KYC — or by card through a partner, and spend per-token.
Related use cases
Start with DeepSeek-V4-Flash
Create an account, top up with crypto, and call DeepSeek-V4-Flash at a fixed per-token rate.
Get startedOpen-weight model served on cost-efficient infrastructure. Model names are trademarks of their respective owners.