大模型排行榜 · hard_prompts

数据来自 LMArena（Chatbot Arena）：全球用户对匿名模型两两盲测投票，按 Elo 机制计算排名——相比跑分基准更接近真实使用体验。本页每周同步官方榜单，当前榜单发布于 2026-07-30。点击模型名可查看其各类别成绩与评分走势。

综合编程创意写作中文英文 exclude_ties expert french german hard_prompts hard_prompts_english industry_business_and_management_and_financial_o industry_entertainment_and_sports_and_media industry_legal_and_government industry_life_and_physical_and_social_science industry_mathematical industry_medicine_and_healthcare industry_software_and_it_services

排名	模型	开发方	Elo 评分	投票数
#1	claude-fable-5	anthropic	1535	—
#2	claude-opus-4-6-thinking	anthropic	1533	—
#3	claude-opus-4-6	anthropic	1526	—
#4	claude-opus-5-high	anthropic	1526	—
#5	claude-opus-4-7-thinking	anthropic	1526	—
#6	claude-opus-5-max	anthropic	1521	—
#7	claude-opus-4-7	anthropic	1518	—
#8	claude-opus-4-8-thinking	anthropic	1514	—
#9	muse-spark-1.1	meta	1510	—
#10	gemini-3.1-pro-preview	google	1507	—
#11	kimi-k3-max	moonshot	1506	—
#12	claude-opus-4-8	anthropic	1505	—
#13	gpt-5.6-sol-xhigh	openai	1505	—
#14	muse-spark	meta	1505	—
#15	gemini-3-pro	google	1504	—
#16	claude-sonnet-4-6	anthropic	1504	—
#17	gemini-3.6-flash	google	1503	—
#18	gpt-5.5-high	openai	1501	—
#19	claude-opus-4-5-20251101-thinking-32k	anthropic	1499	—
#20	gpt-5.4-high	openai	1499	—
#21	claude-opus-4-5-20251101	anthropic	1497	—
#22	gpt-5.2-chat-latest-20260210	openai	1497	—
#23	qwen3.7-max-preview	alibaba	1495	—
#24	gpt-5.5	openai	1494	—
#25	mimo-v2.5-pro	xiaomi	1494	—
#26	gemini-3-flash	google	1493	—
#27	gemini-3.5-flash-high	google	1492	—
#28	qwen3.5-max-preview	alibaba	1492	—
#29	glm-5.1	zai	1492	—
#30	gemini-3.5-flash-medium	google	1492	—
#31	gpt-5.5-instant	openai	1491	—
#32	claude-sonnet-5-high	anthropic	1490	—
#33	gpt-5.6-terra-xhigh	openai	1490	—
#34	glm-5.2-max	zai	1489	—
#35	grok-4.20-beta-0309-reasoning	xai	1489	—
#36	grok-4.5	xai	1488	—
#37	gpt-5.4	openai	1487	—
#38	ernie-5.1	baidu	1487	—
#39	claude-sonnet-4-5-20250929-thinking-32k	anthropic	1487	—
#40	grok-4.20-beta1	xai	1486	—
#41	kimi-k2.6	moonshot	1485	—
#42	grok-4.20-multi-agent-beta-0309	xai	1484	—
#43	claude-sonnet-4-5-20250929	anthropic	1483	—
#44	qwen3.6-max-preview	alibaba	1482	—
#45	claude-opus-4-1-20250805-thinking-16k	anthropic	1480	—
#46	deepseek-v4-pro	deepseek	1480	—
#47	dola-seed-2.0-pro	bytedance	1480	—
#48	glm-5	zai	1478	—
#49	grok-4.1-thinking	xai	1478	—
#50	claude-opus-4-1-20250805	anthropic	1477	—
#51	qwen3.7-plus	alibaba	1477	—
#52	gemini-3.5-flash-lite	google	1477	—
#53	gpt-5.6-luna-xhigh	openai	1477	—
#54	hy3	tencent	1476	—
#55	mimo-v2-pro	xiaomi	1476	—
#56	gemini-3-flash (thinking-minimal)	google	1476	—
#57	grok-4.1	xai	1475	—
#58	gpt-5.1-high	openai	1474	—
#59	deepseek-v4-pro-thinking	deepseek	1474	—
#60	gemma-4-31b	google	1473	—
#61	kimi-k2.5-thinking	moonshot	1472	—
#62	gpt-5.3-chat-latest	openai	1471	—
#63	gpt-5.4-mini-high	openai	1471	—
#64	qwen3.6-plus	alibaba	1467	—
#65	minimax-m3	minimax	1466	—
#66	ernie-5.0-0110	baidu	1465	—
#67	qwen3.5-397b-a17b	alibaba	1464	—
#68	ernie-5.0-preview-1203	baidu	1464	—
#69	inkling	thinky	1463	—
#70	glm-4.7	zai	1463	—
#71	gemma-4-26b-a4b	google	1462	—
#72	mimo-v2.5	xiaomi	1462	—
#73	kimi-k2.5-instant	moonshot	1461	—
#74	gpt-5.2	openai	1460	—
#75	deepseek-v4-flash	deepseek	1460	—
#76	longcat-flash-chat-2602-exp	meituan	1459	—
#77	gpt-5.2-high	openai	1459	—
#78	qwen3-max-preview	alibaba	1459	—
#79	deepseek-v4-flash-thinking	deepseek	1459	—
#80	gemini-2.5-pro	google	1458	—
#81	grok-4.3	xai	1457	—
#82	mimo-v2-omni	xiaomi	1457	—
#83	claude-opus-4-20250514-thinking-16k	anthropic	1456	—
#84	chatgpt-4o-latest-20250326	openai	1456	—
#85	gpt-5.1	openai	1455	—
#86	glm-5v-turbo	zai	1453	—
#87	kimi-k2-thinking-turbo	moonshot	1453	—
#88	gpt-5-chat	openai	1448	—
#89	amazon-nova-experimental-chat-26-02-10	amazon	1448	—
#90	qwen3-235b-a22b-instruct-2507	alibaba	1448	—
#91	qwen3-max-2025-09-23	alibaba	1448	—
#92	deepseek-v3.2-exp	deepseek	1448	—
#93	deepseek-v3.2	deepseek	1447	—
#94	amazon-nova-experimental-chat-26-01-10	amazon	1447	—
#95	deepseek-v3.2-exp-thinking	deepseek	1446	—
#96	gpt-5-high	openai	1446	—
#97	deepseek-v3.2-thinking	deepseek	1446	—
#98	mistral-medium-3.5	mistral	1446	—
#99	deepseek-v3.1-terminus-thinking	deepseek	1446	—
#100	gemini-3.1-flash-lite-preview	google	1445	—
#101	nvidia-nemotron-3-ultra-550b-a55b-nvfp4	nvidia	1443	—
#102	minimax-m2.7	minimax	1442	—
#103	glm-4.6	zai	1442	—
#104	grok-4-1-fast-reasoning	xai	1442	—
#105	gpt-4.5-preview-2025-02-27	openai	1440	—
#106	qwen3-vl-235b-a22b-instruct	alibaba	1440	—
#107	o3-2025-04-16	openai	1440	—
#108	claude-haiku-4-5-20251001	anthropic	1440	—
#109	hunyuan-hy3-preview	tencent	1438	—
#110	deepseek-v3.1-thinking	deepseek	1437	—
#111	kimi-k2-0905-preview	moonshot	1437	—
#112	claude-opus-4-20250514	anthropic	1435	—
#113	deepseek-r1-0528	deepseek	1433	—
#114	grok-4-fast-chat	xai	1433	—
#115	deepseek-v3.1	deepseek	1433	—
#116	qwen3.5-122b-a10b	alibaba	1433	—
#117	glm-4.5	zai	1432	—
#118	mistral-large-3	mistral	1432	—
#119	kimi-k2-0711-preview	moonshot	1431	—
#120	claude-sonnet-4-20250514-thinking-32k	anthropic	1431	—
#121	gpt-4.1-2025-04-14	openai	1431	—
#122	mistral-medium-2508	mistral	1429	—
#123	ernie-5.0-preview-1022	baidu	1428	—
#124	qwen3.5-27b	alibaba	1427	—
#125	longcat-flash-chat	meituan	1426	—
#126	grok-3-preview-02-24	xai	1426	—
#127	deepseek-v3.1-terminus	deepseek	1424	—
#128	gpt-5.4-nano-high	openai	1422	—
#129	qwen3-next-80b-a3b-instruct	alibaba	1421	—
#130	qwen3-235b-a22b-no-thinking	alibaba	1421	—
#131	hunyuan-vision-1.5-thinking	tencent	1420	—
#132	gemini-2.5-flash-preview-09-2025	google	1420	—
#133	grok-4-0709	xai	1420	—
#134	gemini-2.5-flash	google	1420	—
#135	qwen3-vl-235b-a22b-thinking	alibaba	1419	—
#136	deepseek-r1	deepseek	1419	—
#137	claude-sonnet-4-20250514	anthropic	1418	—
#138	o1-2024-12-17	openai	1418	—
#139	qwen3-235b-a22b-thinking-2507	alibaba	1418	—
#140	amazon-nova-experimental-chat-12-10	amazon	1417	—
#141	claude-3-7-sonnet-20250219-thinking-32k	anthropic	1417	—
#142	qwen3.5-flash	alibaba	1416	—
#143	minimax-m2.5	minimax	1415	—
#144	qwen3-coder-480b-a35b-instruct	alibaba	1414	—
#145	mimo-v2-flash (non-thinking)	xiaomi	1414	—
#146	step-3.5-flash	stepfun	1414	—
#147	qwen3.5-35b-a3b	alibaba	1413	—
#148	grok-4-fast-reasoning	xai	1413	—
#149	mimo-v2-flash (thinking)	xiaomi	1412	—
#150	deepseek-v3-0324	deepseek	1409	—
#151	minimax-m2.1-preview	minimax	1408	—
#152	qwen3-30b-a3b-instruct-2507	alibaba	1407	—
#153	o4-mini-2025-04-16	openai	1406	—
#154	trinity-large-preview	—	1404	—
#155	gpt-5-mini-high	openai	1403	—
#156	gpt-4.1-mini-2025-04-14	openai	1402	—
#157	o3-mini-high	openai	1401	—
#158	mistral-medium-2505	mistral	1401	—
#159	hunyuan-t1-20250711	tencent	1399	—
#160	claude-3-5-sonnet-20241022	anthropic	1398	—
#161	claude-3-7-sonnet-20250219	anthropic	1398	—
#162	o1-preview	openai	1396	—
#163	amazon-nova-experimental-chat-11-10	amazon	1393	—
#164	qwen3-235b-a22b	alibaba	1392	—
#165	hunyuan-turbos-20250416	tencent	1391	—
#166	gemini-2.5-flash-lite-preview-09-2025-no-thinking	google	1391	—
#167	glm-4.5-air	zai	1391	—
#168	glm-4.7-flash	zai	1387	—
#169	trinity-large-thinking	—	1386	—
#170	qwen2.5-max	alibaba	1385	—
#171	qwen3-next-80b-a3b-thinking	alibaba	1384	—
#172	glm-4.6v	zai	1383	—
#173	gemini-2.5-flash-lite-preview-06-17-thinking	google	1381	—
#174	minimax-m1	minimax	1381	—
#175	nvidia-nemotron-3-super-120b-a12b	nvidia	1380	—
#176	step-3	stepfun	1378	—
#177	grok-3-mini-high	xai	1377	—
#178	glm-4.5v	zai	1376	—
#179	llama-3.1-nemotron-ultra-253b-v1	nvidia	1375	—
#180	mistral-small-2506	mistral	1374	—
#181	intellect-3	—	1373	—
#182	amazon-nova-experimental-chat-10-20	amazon	1372	—
#183	o3-mini	openai	1370	—
#184	grok-3-mini-beta	xai	1370	—
#185	minimax-m2	minimax	1369	—
#186	command-a-03-2025	cohere	1368	—
#187	hunyuan-turbos-20250226	tencent	1368	—
#188	qwen3-32b	alibaba	1367	—
#189	ling-flash-2.0	ant-group	1366	—
#190	gemma-3-27b-it	google	1365	—
#191	gpt-oss-120b	openai	1362	—
#192	llama-3.3-nemotron-49b-super-v1	nvidia	1361	—
#193	gemini-2.0-flash-001	google	1361	—
#194	nvidia-llama-3.3-nemotron-super-49b-v1.5	nvidia	1361	—
#195	o1-mini	openai	1360	—
#196	claude-3-5-sonnet-20240620	anthropic	1360	—
#197	mercury-2	inception-ai	1360	—
#198	nova-2-lite	amazon	1359	—
#199	amazon-nova-experimental-chat-10-09	amazon	1358	—
#200	qwq-32b	alibaba	1357	—

说明：Elo 评分由 LMArena 根据真实用户盲测投票计算，分差越大代表偏好差距越显著；排名会随每期投票变化。本页仅同步展示并翻译整理，原始数据版权归 LMArena 所有。