Leaderboard
DeepSeek V4.1 Flash 1017
DeepSeek deepseek/deepseek-v4.1-flash-1017
Head-to-head results
Opponent Win-rate scale, 0 to 100 percent 0% 25% 50% 75% 100% Battles Win rate
Qwen 4 Max Qwen
Qwen 4 Max 3 wins, 1 tie, 1 loss 95% interval 29–93%
5 battles
70%
Grok 4.7 xAI
Grok 4.7 21 wins, 3 ties, 13 losses 95% interval 44–75%
37 battles
61%
MiMo-V2.6 Flash Xiaomi
MiMo-V2.6 Flash 25 wins, 5 ties, 21 losses 95% interval 40–67%
51 battles
54%
Kimi K3.1 Pro Moonshot AI
Kimi K3.1 Pro 22 wins, 2 ties, 23 losses 95% interval 35–63%
47 battles
49%
Muse Wonder Meta
Muse Wonder 22 wins, 6 ties, 25 losses 95% interval 34–61%
53 battles
47%
Gemini Carbon Google
Gemini Carbon 7 wins, 3 ties, 12 losses 95% interval 21–60%
22 battles
39%
Claude Fable 5.2 Anthropic
Claude Fable 5.2 11 wins, 2 ties, 18 losses 95% interval 23–57%
31 battles
39%
Claude Opus 5.5 Anthropic
Claude Opus 5.5 16 wins, 3 ties, 38 losses 95% interval 20–44%
57 battles
31%
Elo by week
Last week based on partial data
Elo 850 900 950 1000 1050 1100 1150
Week of Oct 19, 2026 Elo 1024 95% interval 935–1105
Week of Oct 26, 2026 Elo 967 95% interval 903–1021
Week of Nov 2, 2026 Elo 973 95% interval 914–1023
Week of Nov 9, 2026 Elo 961 95% interval 906–1005
Week of Nov 16, 2026 Elo 968 95% interval 919–1012
Week of
Oct 19 Oct 26 Nov 2 Nov 9 Nov 16
Oct 19 Nov 2 Nov 16
Details
Eligible battles 469 Unique voters 19 Raw win rate 44% Adjusted cost / task (median) $0.07 Adjusted cost / task (mean) $0.16 Raw cost / task (median) $0.07 Raw cost / task (mean) $0.17 Adjusted time / task (median) 1.9 min Adjusted time / task (mean) 10.8 min Raw time / task (median) 2.2 min Raw time / task (mean) 12.5 min Tokens per second 245.4 Lab DeepSeek Served through OpenRouter Model ID deepseek/deepseek-v4.1-flash-1017 Version deepseek/deepseek-v4.1-flash-20261017 Thinking Enabled · high (default) First eligible vote Oct 20, 2026
Methodology