Comparing 2 models

DeepSeek V4.1 Flash vs Kimi K3

Benchmarks through

Verdict

DeepSeek V4.1 Flash places higher on 8 of the 10 tasks where both are ranked. DeepSeek V4.1 Flash costs 3.8x less per input token.

  • Choose DeepSeek V4.1 Flash for Value LLM (#1), Structured Data Extraction (#1), JSON & Schema Output (#1).
  • Choose Kimi K3 for Code Completion (#4), Debugging & Code Review (#5).

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 2Kimi K3
Best for
JSON & Schema Output (#1)
JSON & Schema Output (#2)
Provider
deepseek
moonshotai
Context
1.0MBest
1.0MBest
Input / 1M
$0.051Best
$0.19
Output / 1M
$0.20Best
$16
Tool use
yesBest
yesBest
Vision
yesBest
yesBest
Reasoning
yesBest
yesBest
Open-weight
yesBest
yesBest
Released
Sep 10, 2026
Jul 16, 2026

Measured head to head

Benchmarks both models were run on

EvaluationDeepSeek V4.1 FlashKimi K3Source
FrontierMath Tiers 1-3 (Epoch AI)67.4% ± 2.872.2% ± 2.7source · 2026-10-08
GPQA Diamond (Epoch AI)89.8% ± 1.793.1% ± 1.5source · 2026-10-09
LiveBench Agentic Coding77.3%62.2%source · 2026-10-18
LiveBench Coding80.0%81.5%source · 2026-10-18
LiveBench Data Analysis79.3%78.7%source · 2026-10-18
LiveBench Instruction Following70.0%71.4%source · 2026-10-18
LiveBench Language81.2%85.5%source · 2026-10-18
LiveBench Math93.3%84.4%source · 2026-10-18
LiveBench Reasoning86.7%90.7%source · 2026-10-18
LMArena Agent Bash Recovery7.94.4source · 2026-10-08
LMArena Creative Writing1438 Elo1453 Elosource · 2026-11-11
LMArena Instruction Following1473 Elo1484 Elosource · 2026-11-11
LMArena Long Query1474 Elo1490 Elosource · 2026-11-11
LMArena Math1479 Elo1496 Elosource · 2026-11-11
LMArena Text1475 Elo1487 Elosource · 2026-11-11
LMArena Text Factuality1468 Elo1471 Elosource · 2026-11-11
LMArena WebDev1621 Elo1658 Elosource · 2026-11-11
LMArena WebDev Image-to-Web1610 Elo1586 Elosource · 2026-11-10
OTIS Mock AIME (Epoch AI)98.3% ± 0.697.2% ± 1.1source · 2026-10-08
SimpleQA Verified (Epoch AI)42.9% ± 1.650.6% ± 1.6source · 2026-10-13

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