Arena Hard Prompts Elo

Name: Arena Hard Prompts Elo
Creator: KnowAI Space
License: https://www.gnu.org/licenses/gpl-3.0.html

About This Benchmark

Arena Elo score aggregated from user preference votes on hard, complex prompts.

Rank	Model
#1	Anthropic Claude Opus 4.6	1535
#2	Google Gemini 3.1 Pro	1516
#3	ChatGPT GPT-5.4	1507
#4	ChatGPT GPT-5.4 Pro	1507
#5	Grok Grok 4.20	1505
#6	Grok Grok 4.20 (Reasoning)	1505
#7	Anthropic Claude Opus 4.5	1500
#8	Anthropic Claude Sonnet 4.6	1499
#9	Z.ai GLM-5.1	1494
#10	Google Gemini 3 Flash	1493
#11	Anthropic Claude Sonnet 4.5	1485
#12	ChatGPT GPT-5.4 Mini	1484
#13	Anthropic Claude Opus 4.1	1480
#14	Xiaomi MiMo-V2-Pro	1478
#15	Z.ai GLM-5	1476
#16	Moonshot AI Kimi K2.5	1475
#17	Google Gemma 4 31B	1474
#18	Baidu ERNIE 5.0 Thinking	1470
#19	Qwen Qwen3.5 397B A17B	1465
#20	Meituan Longcat Flash Chat	1464
#21	Google Gemini 2.5 Pro	1461
#22	Anthropic Claude Opus 4	1456
#23	Google Gemini 3.1 Flash Lite	1448
#24	DeepSeek DeepSeek V3.2	1447
#25	ChatGPT GPT-5	1446
#26	Grok Grok 4.1 Fast	1442
#27	Grok Grok 4.1 Fast (Reasoning)	1442
#28	Anthropic Claude Haiku 4.5	1437
#29	Anthropic Claude Sonnet 4	1431
#30	MiniMax MiniMax M2.5	1429
#31	MiniMax MiniMax M2.7	1426
#32	Google Gemini 2.5 Flash	1420
#33	ChatGPT GPT-5.4 Nano	1417
#34	ChatGPT GPT-5 Mini	1402
#35	Arcee AI Trinity Large Thinking	1400
#36	Google Gemini 2.5 Flash Lite	1391
#37	ChatGPT GPT OSS 120B	1363
#38	Amazon Nova 2 Lite	1360
#39	ChatGPT GPT-5 Nano	1353
#40	Meta AI Llama 4 Maverick	1338
#41	Meta AI Llama 4 Scout	1329
#42	ChatGPT GPT-4.1	1311