token.app › Compare › DeepSeek V3 vs DeepSeek V4 Flash Latest

DeepSeek V3 vs DeepSeek V4 Flash Latest

DeepSeek V4 Flash Latest is 4.0× cheaper per token on a blended 3:1 basis. DeepSeek V4 Flash Latest accepts the larger context window (1.0M).

Head to head
Blended price uses a 3:1 input:output mix
DeepSeek V3DeepSeek V4 Flash Latest
ProviderDeepSeek~deepseek
Input $/1M$0.257$0.090
Output $/1M$1.03$0.180
Blended $/1M$0.450$0.113
Context window164K1.0M
Max output16K384K
ReleasedDec 2024Aug 2026
LicenceOpen weightsProprietary
Choose DeepSeek V3 if…
  • Open weights — self-hostable
Choose DeepSeek V4 Flash Latest if…
  • Costs less per token — $0.113 vs $0.450 blended
  • Larger context window (1.0M)
  • Newer model (released Aug 2026)
FAQ
Which is better, DeepSeek V3 or DeepSeek V4 Flash Latest?
We do not hold independently-run benchmark scores covering both DeepSeek V3 and DeepSeek V4 Flash Latest, so we make no quality claim. Their pricing and specifications are compared on this page.
Is DeepSeek V3 cheaper than DeepSeek V4 Flash Latest?
DeepSeek V4 Flash Latest is cheaper: $0.113 versus $0.450 per million tokens blended at 3:1 input:output. Input alone: $0.257 vs $0.090. Output alone: $1.03 vs $0.180.
What context windows do DeepSeek V3 and DeepSeek V4 Flash Latest have?
DeepSeek V3: 163,840 tokens. DeepSeek V4 Flash Latest: 1,048,576 tokens.
Who makes DeepSeek V3 and DeepSeek V4 Flash Latest?
DeepSeek V3 is made by DeepSeek. DeepSeek V4 Flash Latest is made by ~deepseek.
Related comparisons