Dear DeepSeek API user,
We plan to raise the overall pricing for DeepSeek API services in the near future, with a significant increase expected. Please plan your usage accordingly. The specific pricing plan will be subject to official notice. Please keep an eye on the Open Platform announcements and check your email for further details. If you continue to use our services after the billing adjustment, you will be deemed to have accepted the adjusted billing terms. If you do not agree, you may choose to cancel your service and apply for a refund. Should you have any questions or require further information, please do not hesitate to contact us.
Thank you for your support and understanding! DeepSeek Team
This is from an email I got.
Couldn't really find reliable information about the price increase because everything about it is paywalled and the Deepseek website doesn't have info on the price increase.
Just use it offline bro if you can. I do not really use LLMs much lately so IDK, but I prefer gemma4 though. Deepseek distilled to 7 or 8b will probably work fine with mid range relatively recent hardware.
hows gemma 4 treating you lately? the prices on openrouter were worse than deepseek at the time so i never even bothered to try it
Quite efficient to run locally, 26b A4B runs pretty well on a 8GB card at around 10 t/s for me. Excells at tasks like translation but falls short at agentic coding in comparison to Qwen.
Gemma 4 26B A4B: Vision is very imprecise. Reasons forever. If it ever stops reasoning then the answers are usually good for a local model, but it also has surprisingly bad failure cases too. The usual. It's very fast.