Gemini 3.8 Flash and DeepSeek V4 Pro are close on price. Google charges $0.75 per million input tokens and $3.75 per million output tokens for 3.8 Flash until the end of 2026, then $1.50 and $7.50. DeepSeek charges $1.32 and $3.96 for V4 Pro at peak times and half that off-peak. Both have a context window of about 1M tokens.
The bigger differences are input types and control. Gemini 3.8 Flash accepts text, images, video, audio and PDFs and can ground answers in Google Search. V4 Pro is text-only on DeepSeek's API. On the other hand, V4 Pro's weights are on Hugging Face under the MIT licence, so you can run it on your own hardware, while Gemini models are only available as a Google service. DeepSeek's newer V4.1-Flash adds image input at $0.30 and $1.20. Compare them on your own prompts in Omny.
Which should you choose?
Gemini
Gemini 3.8 Flash · 1M context
Pick it if
you need video, audio or image input, or Gemini inside Google's apps.
DeepSeek
DeepSeek V4 Pro 0423 · 1M context
Pick it if
you want MIT-licensed weights you can self-host.
Pick Gemini 3.8 Flash for multimodal input, Google Search grounding and Google's apps. Pick DeepSeek V4 Pro for open weights and self-hosting, or V4.1-Flash for the lowest API price. Both are inexpensive enough to test side by side.
Questions
Is DeepSeek multimodal like Gemini?
Partly. DeepSeek V4 Pro does not accept images on DeepSeek's API, but V4.1-Flash, released on September 10, 2026, adds native image input. Gemini 3.8 Flash accepts text, images, video, audio and PDFs.
Which is cheaper?
It depends on the model and the time of day. At peak rates Gemini 3.8 Flash ($0.75 in, $3.75 out) undercuts DeepSeek V4 Pro ($1.32 in, $3.96 out). DeepSeek's off-peak rates are half its peak rates, and V4.1-Flash is cheaper than both at $0.30 and $1.20 at peak.