API prices / Compare

DeepSeek V4 Pro vs Gemini 3.8 Flash: price

Gemini 3.8 Flash is 1.3× cheaper than DeepSeek V4 Pro for a typical chat workload (1,500 tokens in, 500 out).

ItemDeepSeek V4 ProGemini 3.8 Flash
Input per 1M tokens$1.32$0.75
Output per 1M tokens$3.96$3.75
Cached input per 1M$0.044$0.075
1,000 chat messages1,500 tokens in, 500 out each$3.96$3.00
Summarize a 50-page PDF30,000 tokens in, 1,000 out$0.044$0.026
One coding-agent task60,000 tokens in, 4,000 out, no caching$0.095$0.06

DeepSeek V4 Pro: Peak-hour price (weekdays 01:00–04:00 and 06:00–10:00 UTC). Off-peak: $0.66 in / $1.98 out.

Gemini 3.8 Flash: Promo price through Dec 31, 2026; $1.50 in / $7.50 out from Jan 1, 2027.

List prices checked September 28, 2026. Price is half the story: see how people rate each one today.

DeepSeek V4 Pro vs Gemini 3.8 Flash in numbers

Output, usually the bigger part of the bill, costs $3.96 on DeepSeek V4 Pro and $3.75 on Gemini 3.8 Flash, so Gemini 3.8 Flash is 1.1× cheaper there. Input costs $1.32 versus $0.75 per million tokens. For a coding-agent task (60,000 tokens in, 4,000 out) that is $0.095 on DeepSeek V4 Pro and $0.06 on Gemini 3.8 Flash. DeepSeek V4 Pro accepts 1M tokens of context and Gemini 3.8 Flash 1.05M tokens. DeepSeek V4 Pro came out on August 13, 2026 and Gemini 3.8 Flash on September 2, 2026.

DeepSeek V4 Pro is made by DeepSeek; Gemini 3.8 Flash by Google. Estimate your own bill for both in the cost calculator.

Which one feels dumber today?

  1. –no votes yet
  2. –no votes yet

Frequently asked questions

Is DeepSeek V4 Pro or Gemini 3.8 Flash cheaper?

Gemini 3.8 Flash is 1.3× cheaper than DeepSeek V4 Pro for a typical chat workload (1,500 tokens in, 500 out). Output, usually the bigger part of the bill, costs $3.96 on DeepSeek V4 Pro and $3.75 on Gemini 3.8 Flash, so Gemini 3.8 Flash is 1.1× cheaper there.

Which is better, DeepSeek V4 Pro or Gemini 3.8 Flash?

It depends on the task, and price is only part of it. Check how people rate each one today on the DeepSeek V4 Pro and Gemini 3.8 Flash pages, and read what users report in Discussions.