Comparing 2 models

DeepSeek V4.1 Flash vs Gemini 3.8 Flash

Benchmarks through

Verdict

DeepSeek V4.1 Flash places higher on 3 of the 4 tasks where both are ranked. DeepSeek V4.1 Flash costs 16.3x less per input token. Only DeepSeek V4.1 Flash has open weights for self-hosting.

  • Choose DeepSeek V4.1 Flash for Value LLM (#1), Cheapest Capable LLM (#2), Cheap LLM for High Volume (#2), you need to self-host.
  • Choose Gemini 3.8 Flash for Fast Summaries (#1).

Built from the published task rankings below; ranks come from each task's own weighted benchmarks, so a win on one task says nothing about another.

metric
Model 1DeepSeek V4.1 Flash
Model 2Gemini 3.8 Flash
Best for
JSON & Schema Output (#1)
Business Writing (#1)
Provider
deepseek
google
Context
1.0MBest
1.0MBest
Input / 1M
$0.051Best
$0.83
Output / 1M
$0.20Best
$4.13
Tool use
yesBest
yesBest
Vision
yesBest
yesBest
Reasoning
yesBest
yesBest
Open-weight
yesBest
—
Released
Sep 10, 2026
Sep 2, 2026

Measured head to head

Benchmarks both models were run on

EvaluationDeepSeek V4.1 FlashGemini 3.8 FlashSource
FrontierMath Tiers 1-3 (Epoch AI)67.4% ± 2.868.4% ± 2.8source · 2026-10-08
GPQA Diamond (Epoch AI)89.8% ± 1.795.4% ± 1.4source · 2026-10-09
LiveBench Agentic Coding77.3%54.2%source · 2026-10-18
LiveBench Coding80.0%72.5%source · 2026-10-18
LiveBench Data Analysis79.3%54.0%source · 2026-10-18
LiveBench Instruction Following70.0%81.4%source · 2026-10-18
LiveBench Language81.2%87.8%source · 2026-10-18
LiveBench Math93.3%91.6%source · 2026-10-18
LiveBench Reasoning86.7%89.3%source · 2026-10-18
LMArena Agent Bash Recovery7.9-0.7source · 2026-10-08
LMArena Creative Writing1438 Elo1493 Elosource · 2026-11-11
LMArena Instruction Following1473 Elo1493 Elosource · 2026-11-11
LMArena Long Query1474 Elo1508 Elosource · 2026-11-11
LMArena Math1479 Elo1525 Elosource · 2026-11-11
LMArena Text1475 Elo1497 Elosource · 2026-11-11
LMArena Text Factuality1468 Elo1470 Elosource · 2026-11-11
LMArena Vision1263 Elo1293 Elosource · 2026-11-09
LMArena Vision Diagrams1296 Elo1342 Elosource · 2026-11-09
LMArena Vision OCR1274 Elo1315 Elosource · 2026-11-09
LMArena WebDev1621 Elo1583 Elosource · 2026-11-11
LMArena WebDev Image-to-Web1610 Elo1594 Elosource · 2026-11-10
OTIS Mock AIME (Epoch AI)98.3% ± 0.698.9% ± 0.9source · 2026-10-08
SimpleQA Verified (Epoch AI)42.9% ± 1.669.7% ± 1.5source · 2026-10-13

Values are as published by each evaluator, in its own units; rows from different evaluators are not on a common scale.