大模型排行榜 · hard_prompts_english

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-opus-4-6-thinking	anthropic	1538	—
#2	claude-fable-5	anthropic	1534	—
#3	claude-opus-4-6	anthropic	1530	—
#4	claude-opus-4-7-thinking	anthropic	1526	—
#5	claude-opus-5-high	anthropic	1524	—
#6	claude-opus-4-7	anthropic	1523	—
#7	claude-opus-4-8-thinking	anthropic	1514	—
#8	gpt-5.6-sol-xhigh	openai	1510	—
#9	claude-opus-4-8	anthropic	1509	—
#10	muse-spark-1.1	meta	1508	—
#11	claude-sonnet-4-6	anthropic	1508	—
#12	kimi-k3-max	moonshot	1508	—
#13	muse-spark	meta	1507	—
#14	gemini-3.1-pro-preview	google	1507	—
#15	claude-opus-4-5-20251101-thinking-32k	anthropic	1505	—
#16	mimo-v2.5-pro	xiaomi	1503	—
#17	claude-opus-5-max	anthropic	1502	—
#18	glm-5.1	zai	1502	—
#19	gemini-3.6-flash	google	1502	—
#20	gemini-3-pro	google	1502	—
#21	gpt-5.5-high	openai	1500	—
#22	gpt-5.2-chat-latest-20260210	openai	1499	—
#23	claude-opus-4-5-20251101	anthropic	1498	—
#24	claude-sonnet-4-5-20250929-thinking-32k	anthropic	1497	—
#25	gpt-5.4-high	openai	1495	—
#26	qwen3.5-max-preview	alibaba	1494	—
#27	gpt-5.5	openai	1494	—
#28	claude-sonnet-5-high	anthropic	1493	—
#29	ernie-5.1	baidu	1492	—
#30	gemini-3.5-flash-medium	google	1491	—
#31	claude-sonnet-4-5-20250929	anthropic	1490	—
#32	qwen3.7-max-preview	alibaba	1490	—
#33	grok-4.20-beta-0309-reasoning	xai	1490	—
#34	gpt-5.6-luna-xhigh	openai	1490	—
#35	mimo-v2-pro	xiaomi	1489	—
#36	claude-opus-4-1-20250805-thinking-16k	anthropic	1489	—
#37	gpt-5.5-instant	openai	1488	—
#38	glm-5.2-max	zai	1488	—
#39	grok-4.5	xai	1488	—
#40	gemini-3-flash	google	1487	—
#41	grok-4.20-beta1	xai	1487	—
#42	gpt-5.4	openai	1487	—
#43	kimi-k2.6	moonshot	1487	—
#44	glm-5	zai	1487	—
#45	gemini-3.5-flash-high	google	1486	—
#46	gpt-5.6-terra-xhigh	openai	1484	—
#47	deepseek-v4-pro	deepseek	1484	—
#48	qwen3.6-max-preview	alibaba	1484	—
#49	gemma-4-31b	google	1483	—
#50	claude-opus-4-1-20250805	anthropic	1482	—
#51	grok-4.20-multi-agent-beta-0309	xai	1481	—
#52	deepseek-v4-pro-thinking	deepseek	1481	—
#53	hy3	tencent	1481	—
#54	dola-seed-2.0-pro	bytedance	1480	—
#55	qwen3.7-plus	alibaba	1479	—
#56	gpt-5.1-high	openai	1479	—
#57	grok-4.1-thinking	xai	1479	—
#58	kimi-k2.5-thinking	moonshot	1478	—
#59	grok-4.1	xai	1478	—
#60	longcat-flash-chat-2602-exp	meituan	1476	—
#61	qwen3.6-plus	alibaba	1476	—
#62	minimax-m3	minimax	1475	—
#63	glm-4.7	zai	1473	—
#64	gemini-3-flash (thinking-minimal)	google	1472	—
#65	mimo-v2.5	xiaomi	1472	—
#66	gemini-3.5-flash-lite	google	1472	—
#67	gemma-4-26b-a4b	google	1469	—
#68	gpt-5.3-chat-latest	openai	1469	—
#69	glm-5v-turbo	zai	1468	—
#70	gpt-5.4-mini-high	openai	1468	—
#71	amazon-nova-experimental-chat-26-01-10	amazon	1467	—
#72	qwen3.5-397b-a17b	alibaba	1467	—
#73	ernie-5.0-0110	baidu	1467	—
#74	inkling	thinky	1466	—
#75	kimi-k2.5-instant	moonshot	1465	—
#76	mimo-v2-omni	xiaomi	1465	—
#77	deepseek-v4-flash-thinking	deepseek	1465	—
#78	deepseek-v4-flash	deepseek	1464	—
#79	deepseek-v3.2-exp-thinking	deepseek	1464	—
#80	claude-opus-4-20250514-thinking-16k	anthropic	1463	—
#81	deepseek-v3.2-thinking	deepseek	1462	—
#82	grok-4.3	xai	1461	—
#83	gpt-5.2-high	openai	1461	—
#84	ernie-5.0-preview-1203	baidu	1460	—
#85	kimi-k2-thinking-turbo	moonshot	1460	—
#86	minimax-m2.7	minimax	1459	—
#87	gpt-5.2	openai	1459	—
#88	mistral-medium-3.5	mistral	1458	—
#89	qwen3-max-2025-09-23	alibaba	1458	—
#90	gpt-5.1	openai	1458	—
#91	qwen3-max-preview	alibaba	1457	—
#92	gemini-2.5-pro	google	1457	—
#93	deepseek-v3.2	deepseek	1455	—
#94	claude-haiku-4-5-20251001	anthropic	1455	—
#95	chatgpt-4o-latest-20250326	openai	1455	—
#96	amazon-nova-experimental-chat-26-02-10	amazon	1454	—
#97	deepseek-v3.2-exp	deepseek	1454	—
#98	gpt-5-high	openai	1453	—
#99	qwen3-vl-235b-a22b-instruct	alibaba	1452	—
#100	qwen3-235b-a22b-instruct-2507	alibaba	1451	—
#101	deepseek-v3.1-terminus-thinking	deepseek	1450	—
#102	claude-sonnet-4-20250514-thinking-32k	anthropic	1448	—
#103	glm-4.6	zai	1448	—
#104	gpt-5-chat	openai	1448	—
#105	nvidia-nemotron-3-ultra-550b-a55b-nvfp4	nvidia	1446	—
#106	o3-2025-04-16	openai	1446	—
#107	grok-4-fast-chat	xai	1445	—
#108	deepseek-v3.1-thinking	deepseek	1445	—
#109	grok-4-1-fast-reasoning	xai	1444	—
#110	gpt-4.5-preview-2025-02-27	openai	1444	—
#111	qwen3.5-122b-a10b	alibaba	1444	—
#112	claude-opus-4-20250514	anthropic	1443	—
#113	gemini-3.1-flash-lite-preview	google	1442	—
#114	amazon-nova-experimental-chat-12-10	amazon	1441	—
#115	qwen3.5-27b	alibaba	1441	—
#116	mistral-large-3	mistral	1440	—
#117	longcat-flash-chat	meituan	1439	—
#118	kimi-k2-0905-preview	moonshot	1439	—
#119	mistral-medium-2508	mistral	1439	—
#120	deepseek-r1-0528	deepseek	1438	—
#121	hunyuan-hy3-preview	tencent	1438	—
#122	gpt-4.1-2025-04-14	openai	1437	—
#123	deepseek-v3.1	deepseek	1435	—
#124	mimo-v2-flash (non-thinking)	xiaomi	1434	—
#125	kimi-k2-0711-preview	moonshot	1434	—
#126	deepseek-r1	deepseek	1433	—
#127	grok-3-preview-02-24	xai	1433	—
#128	glm-4.5	zai	1432	—
#129	gpt-5.4-nano-high	openai	1431	—
#130	minimax-m2.5	minimax	1430	—
#131	qwen3-next-80b-a3b-instruct	alibaba	1429	—
#132	claude-3-7-sonnet-20250219-thinking-32k	anthropic	1429	—
#133	deepseek-v3.1-terminus	deepseek	1428	—
#134	qwen3-coder-480b-a35b-instruct	alibaba	1428	—
#135	qwen3-235b-a22b-thinking-2507	alibaba	1428	—
#136	ernie-5.0-preview-1022	baidu	1427	—
#137	step-3.5-flash	stepfun	1425	—
#138	grok-4-0709	xai	1425	—
#139	claude-sonnet-4-20250514	anthropic	1425	—
#140	minimax-m2.1-preview	minimax	1424	—
#141	hunyuan-vision-1.5-thinking	tencent	1424	—
#142	mimo-v2-flash (thinking)	xiaomi	1424	—
#143	qwen3.5-35b-a3b	alibaba	1424	—
#144	grok-4-fast-reasoning	xai	1424	—
#145	qwen3.5-flash	alibaba	1423	—
#146	o1-2024-12-17	openai	1423	—
#147	gemini-2.5-flash-preview-09-2025	google	1423	—
#148	qwen3-235b-a22b-no-thinking	alibaba	1420	—
#149	qwen3-vl-235b-a22b-thinking	alibaba	1419	—
#150	gemini-2.5-flash	google	1418	—
#151	deepseek-v3-0324	deepseek	1418	—
#152	trinity-large-preview	—	1418	—
#153	o4-mini-2025-04-16	openai	1417	—
#154	qwen3-30b-a3b-instruct-2507	alibaba	1416	—
#155	o3-mini-high	openai	1414	—
#156	o1-preview	openai	1413	—
#157	gpt-4.1-mini-2025-04-14	openai	1413	—
#158	mistral-medium-2505	mistral	1412	—
#159	gpt-5-mini-high	openai	1410	—
#160	claude-3-7-sonnet-20250219	anthropic	1406	—
#161	claude-3-5-sonnet-20241022	anthropic	1405	—
#162	glm-4.6v	zai	1405	—
#163	glm-4.7-flash	zai	1404	—
#164	glm-4.5-air	zai	1403	—
#165	qwen3-235b-a22b	alibaba	1403	—
#166	amazon-nova-experimental-chat-11-10	amazon	1403	—
#167	step-3	stepfun	1402	—
#168	intellect-3	—	1399	—
#169	hunyuan-t1-20250711	tencent	1398	—
#170	mistral-small-2506	mistral	1397	—
#171	llama-3.1-nemotron-ultra-253b-v1	nvidia	1397	—
#172	glm-4.5v	zai	1397	—
#173	gemini-2.5-flash-lite-preview-09-2025-no-thinking	google	1397	—
#174	qwen3-next-80b-a3b-thinking	alibaba	1396	—
#175	trinity-large-thinking	—	1395	—
#176	nvidia-nemotron-3-super-120b-a12b	nvidia	1395	—
#177	amazon-nova-experimental-chat-10-20	amazon	1395	—
#178	minimax-m1	minimax	1395	—
#179	hunyuan-turbos-20250416	tencent	1391	—
#180	ling-flash-2.0	ant-group	1389	—
#181	minimax-m2	minimax	1388	—
#182	qwen2.5-max	alibaba	1388	—
#183	llama-3.3-nemotron-49b-super-v1	nvidia	1387	—
#184	gemini-2.5-flash-lite-preview-06-17-thinking	google	1383	—
#185	o3-mini	openai	1383	—
#186	hunyuan-turbos-20250226	tencent	1382	—
#187	grok-3-mini-high	xai	1382	—
#188	qwen3-32b	alibaba	1382	—
#189	grok-3-mini-beta	xai	1382	—
#190	ring-flash-2.0	ant-group	1379	—
#191	nova-2-lite	amazon	1378	—
#192	command-a-03-2025	cohere	1376	—
#193	o1-mini	openai	1375	—
#194	olmo-3.1-32b-instruct	allenai	1373	—
#195	qwq-32b	alibaba	1373	—
#196	mercury-2	inception-ai	1372	—
#197	gemma-3-27b-it	google	1370	—
#198	hunyuan-turbo-0110	tencent	1369	—
#199	amazon-nova-experimental-chat-10-09	amazon	1369	—
#200	gpt-5-nano-high	openai	1368	—

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。