Skip to content

DeepSeek V4.1-Flash vs Gemini 3.8 Flash

DeepSeek V4.1-Flash is from DeepSeek, released September 10, 2026. Gemini 3.8 Flash is from Google, released September 2, 2026.

  • LMArena Elo1473 vs 1494. Gemini 3.8 Flash is 1.4% ahead on LMArena Elo.
  • Input per 1M$0.30 vs $0.75. DeepSeek V4.1-Flash costs 60% less per million input tokens.
  • Output per 1M$1.2 vs $3.75. DeepSeek V4.1-Flash costs 68% less per million output tokens.
  • Context window1M vs 1M. Both hold 1M tokens of context.

Prices are the vendors’ published API rates; Elo is LMArena’s text board. Each figure is dated on its model page.

Interactive Comparison

Loading chart...
$0.30 / $1.20 per 1M tokens (peak, cache miss); $0.15 / $0.60 off-peak
$0.75 / $3.75 per 1M tokens (introductory, to 31 Dec 2026)
Identity
DeveloperDeepSeekGoogle
ReleasedSep 2026Sep 2026
StatusGAGA
LicenceMITProprietary
Self-hostableYes - open weights on Hugging FaceNo
Cost
Blended $/1M tokensinput × 0.75 + output × 0.25$0.525 / 1M tokens$1.500 / 1M tokens
Input price$0.300 / 1M tokens$0.750 / 1M tokens
Output price$1.200 / 1M tokens$3.750 / 1M tokens
Cached input$0.006 / 1M tokens (peak); $0.003 off-peak—
Batch discountOff-peak billing: 50% off all rates—
Capacity
Context window1M tokens1M tokens
Max output384k tokens64k tokens
Long-context surchargeNone (flat rate across the 1M window)—
Capability
Vision inYes—
Audio inNo—
Function callingYes—
Structured outputYes - JSON output—
Extended reasoningYes - thinking (default) and non-thinking modesTunable thinking: low, medium (default), high. The "minimal" level is not supported.
Web searchNo—
Code executionNo—
Access
APIYesYes
Chat appYesYes
Cloud marketplacesNot offered—
Fine-tuningNot offered (weights are open)—
Measured quality
LMArena Elo1473 (checked Oct 2026)1494 (checked Oct 2026)
LMArena rankRank 40Rank 11
GPQA Diamond—95
Elo per dollarLMArena Elo ÷ blended $/1M tokens2806996

Add a third model to this comparison