GLOBAL DATA · THIRD-PARTY INDEX

AI model rankings.

One view of model capability. All the context to make it useful.

Email following opens soon

METRICEpoch Capabilities Index

FIRST OBSERVED / SNAPSHOT2026-11-09

LAST SOURCE CHECK (UTC)2026-11-09T22:32

COVERAGEGlobal · 255 matching models

Source note ECI is Epoch AI’s composite capability index, not a RankMondo evaluation. Uncertainty intervals may overlap; small score differences may not be meaningful. Bayesian ECI, introduced by Epoch AI on 4 November 2026. Earlier-method snapshots are not directly compared. Source methodology changelog. Source publication timestamp is unavailable. Model release dates are not dataset dates.

Snapshot observed 2026-11-09 · ECI, not RankMondo testingSource: Epoch AI ↗ · CC BY 4.0

Select a model in the chart to see its score and source interval.

THE LEADERBOARD

Capability, with context

Rank	Model / Organisation	ECI score ↓	90% credible interval	Change	Access
01	Claude Fable 5.2Anthropic	174.24	170.12–178.79	New	API access
02	GPT-6 AstraOpenAI	169.16	166.19–172.51	↓ 1	API access
03	GPT-6.1 SolOpenAI	168.87	165.90–172.24	↓ 1	API access
04	Claude Opus 5.5Anthropic	168.67	165.78–171.91	↓ 1	API access
05	GPT-6.1 TerraOpenAI	168.11	165.09–171.50	↓ 1	API access
06	Claude Fable 5.1Anthropic	167.04	164.29–170.20	↓ 1	API access
07	Gemini 4 ArgonGoogle DeepMind	166.53	163.88–169.53	↓ 1	Limited access
08	Claude Sonnet 5.5Anthropic	166.32	163.65–169.35	↓ 1	API access
09	Claude Opus 5Anthropic	164.48	162.02–167.30	↓ 1	API access
10	Claude Fable 5Anthropic	164.25	161.83–167.02	↓ 1	API access
11	GPT-6 SolOpenAI	164.00	161.51–166.81	↓ 1	API access
12	GPT-5.6 SolOpenAI	163.02	160.72–165.59	↓ 1	API access
13	GPT-5.5 ProOpenAI	162.75	159.99–165.79	↓ 1	API access
14	Kimi K3.1 ProMoonshot	162.02	159.45–164.95	New	Open weights (restricted use)
15	Muse WonderMeta AI	162.01	159.60–164.67	↓ 2	API access
16	GPT-5.6 TerraOpenAI	160.86	158.62–163.30	↓ 2	API access
17	Claude Haiku 5.5Anthropic	160.25	158.02–162.73	↓ 2	API access
18	GPT-5.4 ProOpenAI	160.17	157.72–162.87	↓ 2	API access
19	GPT-5.5OpenAI	160.06	158.02–162.38	↓ 2	API access
20	Claude Opus 4.8Anthropic	159.91	157.82–162.25	↓ 2	API access
21	DeepSeek V4.1 Flash 1017DeepSeek	159.34	157.08–161.86	↓ 2	Open weights (unrestricted)
22	Gemini 3.8 FlashGoogle DeepMind	158.66	156.59–160.92	↓ 2	API access
23	Kimi K3Moonshot	158.53	156.54–160.71	↓ 2	Open weights (non-commercial)
24	Gemini 3.7 FlashGoogle DeepMind	158.44	156.44–160.62	↓ 2	API access
25	Muse Spark 1.3Meta AI	157.97	155.86–160.26	↓ 2	API access
26	Grok 4.6xAI	157.91	155.87–160.12	↓ 2	API access
27	GPT-5.4OpenAI	157.81	155.95–159.96	↓ 2	API access
28	Claude Opus 4.7Anthropic	157.56	155.69–159.63	↓ 2	API access
29	Qwen 3.8 MaxAlibaba	157.45	155.35–159.70	↓ 2	API access
30	GPT-5.2 ProOpenAI	157.21	154.95–159.78	↓ 2	API access
31	GLM-5.3Z.ai (Zhipu AI)	156.94	154.83–159.24	↓ 2	Open weights (restricted use)
32	GPT-5.6 LunaOpenAI	156.75	154.87–158.80	↓ 2	API access
33	Claude Sonnet 5Anthropic	156.72	154.73–158.86	↓ 2	API access
34	GPT-6 LunaOpenAI	156.50	154.60–158.57	↓ 2	API access
35	DeepSeek V4.1 FlashDeepSeek	156.45	154.55–158.55	↓ 2	Open weights (unrestricted)
36	Claude Opus 4.6Anthropic	156.31	154.57–158.22	↓ 2	API access
37	Gemini 3.1 ProGoogle DeepMind	156.25	154.51–158.14	↓ 2	API access
38	Grok 4.7xAI	156.05	154.15–158.13	↓ 2	API access
39	DeepSeek V4 Pro 0813DeepSeek	155.90	154.03–158.01	↓ 2	Open weights (unrestricted)
40	Qwen3.8 Max (0902)Alibaba	155.83	153.62–158.18	↓ 2	API access
41	Muse Spark OpenMeta AI	155.41	153.36–157.68	↓ 2	Open weights (restricted use)
42	Gemini 3.5 FlashGoogle DeepMind	155.22	153.40–157.24	↓ 2	API access
43	Gemini 3.6 FlashGoogle DeepMind	155.20	153.32–157.17	↓ 2	API access
44	Grok 4.5xAI	155.08	153.25–157.09	↓ 2	API access
45	Muse Spark 1.2Meta AI	155.00	152.71–157.49	↓ 2	API access
46	Muse Spark 1.1Meta AI	154.85	152.25–157.56	↓ 2	API access
47	Mistral Large 4Mistral AI	154.59	151.79–157.46	↓ 2	Open weights (restricted use)
48	Qwen3.7-MaxAlibaba	154.50	152.45–156.58	↓ 2	API access
49	GPT-5.2OpenAI	154.40	152.77–156.24	↓ 2	API access
50	DeepSeek V4 Flash 0731DeepSeek	154.39	152.51–156.38	↓ 2	Open weights (unrestricted)
51	GLM-5.3-FlashZ.ai (Zhipu AI)	153.60	151.81–155.58	↓ 2	Open weights (unrestricted)
52	Muse SparkMeta AI	153.12	150.98–155.39	↓ 2	API access
53	GLM-5.2Z.ai (Zhipu AI)	153.01	151.25–154.93	↓ 2	Open weights (unrestricted)
54	Gemini 3 ProGoogle DeepMind	152.89	151.34–154.57	↓ 2	API access
55	Claude Sonnet 4.6Anthropic	152.84	151.18–154.58	↓ 2	API access
56	Qwen 3.8 27BAlibaba	152.70	150.70–154.92	↓ 2	Open weights (unrestricted)
57	Mistral Large 4 PreviewMistral AI	152.55	150.67–154.49	↓ 2	Open weights (restricted use)
58	Gemini 3 FlashGoogle DeepMind	152.29	150.73–153.99	↓ 1	API access
59	Grok 4.20xAI	152.27	150.59–154.09	↓ 3	API access
60	Claude Opus 4.5Anthropic	151.83	150.28–153.47	↓ 2	API access
61	Kimi K2.6Moonshot	151.54	149.75–153.38	↓ 2	Open weights (unrestricted)
62	GPT-5 ProOpenAI	151.17	149.26–153.17	↓ 1	API access
63	Kimi K2.7 CodeMoonshot	151.14	149.14–153.23	↓ 3	Open weights (unrestricted)
64	GPT-5.1OpenAI	150.89	149.38–152.52	↓ 2	API access
65	Beam 1.0Reflection	150.81	148.70–153.02	↓ 1	Open weights (unrestricted)
66	Inkling-SmallThinking Machines	150.76	148.99–152.62	↓ 3	Open weights (unrestricted)
67	GLM-5.1Z.ai (Zhipu AI)	150.72	148.70–152.83	↓ 2	Open weights (unrestricted)
68	DeepSeek-V4-ProDeepSeek	150.14	148.39–151.92	↓ 2	Open weights (unrestricted)
69	GPT-5OpenAI	150.00	Not supplied	↓ 2	API access
70	Grok 4.3 BetaxAI	149.97	147.97–152.02	↓ 3	Hosted access (no API)
71	Qwen 3.6 Max (Preview)Alibaba	149.80	147.79–151.91	↓ 2	Unknown
72	GPT-5.4 MiniOpenAI	149.33	147.69–151.11	↓ 2	API access
73	Qwen 3.6 PlusAlibaba	148.89	146.93–150.89	↓ 2	API access
74	Kimi K2.5Moonshot	148.75	147.26–150.29	↓ 2	Open weights (unrestricted)
75	InklingThinking Machines	148.36	146.82–149.90	↓ 2	Open weights (unrestricted)
76	Qwen3.5 397B-A17BAlibaba	147.91	145.79–150.17	↓ 2	Open weights (unrestricted)
77	Qwen3.7-PlusAlibaba	147.87	145.90–150.02	↓ 2	API access
78	MiniMax-M3MiniMax	147.77	145.56–149.97	↓ 2	Open weights (restricted use)
79	Claude Sonnet 4.5Anthropic	147.60	146.16–149.05	↓ 2	API access
80	o3-proOpenAI	147.45	145.58–149.31	↓ 2	API access
81	Qwen 3.5 Plus (hosted 397B-A17B)Alibaba	147.28	145.46–149.12	↓ 2	Unknown
82	Nemotron 3 UltraNvidia	147.11	144.99–149.29	↓ 2	Open weights (unrestricted)
83	o3OpenAI	147.01	145.54–148.54	↓ 2	API access
84	Grok 4xAI	146.88	145.46–148.39	↓ 2	API access
85	DeepSeek-V4-FlashDeepSeek	146.87	143.63–150.17	↓ 3	Open weights (unrestricted)
86	Qwen3.6 27BAlibaba	146.72	144.62–148.86	↓ 2	Open weights (unrestricted)
87	GLM-5Z.ai (Zhipu AI)	146.47	144.90–148.01	↓ 2	Open weights (unrestricted)
88	GPT-5.4 NanoOpenAI	146.39	144.73–148.04	↓ 2	API access
89	DeepSeek-V3.2DeepSeek	146.24	144.80–147.74	↓ 2	Open weights (unrestricted)
90	GPT-5 miniOpenAI	146.08	144.51–147.60	↓ 2	API access
91	Gemini 3.5 Flash-LiteGoogle DeepMind	145.58	143.97–147.21	↓ 2	API access
92	Kimi K2 ThinkingMoonshot	145.53	143.65–147.36	↓ 2	Open weights (restricted use)
93	o4-miniOpenAI	145.22	143.77–146.65	↓ 2	API access
94	Gemini 3.1 Flash-LiteGoogle	144.90	142.73–147.08	↓ 2	API access
95	Gemini 2.5 Pro (Jun 2025)Google DeepMind	144.81	143.36–146.22	↓ 2	API access
96	Claude Opus 4.1Anthropic	144.80	143.03–146.53	↓ 2	API access
97	DeepSeek-V3.2-ExpDeepSeek	144.41	142.11–146.67	↓ 2	Open weights (unrestricted)
98	Grok 4 FastxAI	144.33	142.21–146.33	↓ 2	API access
99	Gemma 4 31B ITGoogle DeepMind	144.24	141.77–146.81	↓ 2	Open weights (restricted use)
100	Qwen 3.6 35B-A3BAlibaba	144.05	142.13–145.93	↓ 2	Open weights (unrestricted)
101	Qwen3.5-35B-A3BAlibaba	143.97	141.46–146.58	↓ 2	Open weights (unrestricted)
102	Gemini 2.5 Pro (Mar 2025)Google DeepMind	143.91	141.99–145.92	↓ 2	API access
103	Qwen 3.5 Flash (hosted 35B-A3B)Alibaba	143.81	141.84–145.77	↓ 2	Unknown
104	Qwen 3.6 FlashAlibaba	143.68	141.75–145.62	↓ 2	Unknown
105	GLM-4.7Z.ai (Zhipu AI)	143.27	141.23–145.19	↓ 2	Open weights (unrestricted)
106	Qwen3-235B-A22B-Thinking (Jul 2025)Alibaba	143.24	141.32–145.13	↓ 2	Open weights (unrestricted)
107	Claude Haiku 4.5Anthropic	142.86	141.17–144.46	↓ 1	API access
107	Claude Opus 4Anthropic	142.86	141.26–144.48	↓ 2	API access
109	Qwen3-MaxAlibaba	142.58	140.46–144.68	↓ 2	API access
110	Gemini 2.5 Flash (Sep 2025)Google DeepMind	141.92	139.25–144.48	↓ 2	API access
111	Claude Sonnet 4Anthropic	141.83	140.32–143.34	↓ 2	API access
112	Gemma 4 26B A4BGoogle DeepMind	141.70	139.15–144.20	↓ 2	Open weights (unrestricted)
113	Mistral Medium 3.5Mistral AI	141.60	138.28–144.74	↓ 2	Open weights (restricted use)
114	Gemini 2.5 Flash (May 2025)Google DeepMind	141.59	139.37–143.71	↓ 2	API access
115	Kimi K2 (Jul 2025)Moonshot	141.32	138.40–144.02	↓ 2	Open weights (restricted use)
116	o1OpenAI	141.16	139.51–142.77	↓ 2	API access
117	Claude 3.7 SonnetAnthropic	141.00	139.45–142.52	↓ 2	API access
118	Gemini 2.5 Pro (May 2025)Google DeepMind	140.91	138.61–143.27	↓ 2	API access
119	DeepSeek-V3.1DeepSeek	140.84	137.71–143.99	↓ 2	Open weights (unrestricted)
120	Gemini 2.5 Flash (Jun 2025)Google DeepMind	140.83	138.44–143.10	↓ 2	API access
121	DeepSeek-R1 (May 2025)DeepSeek	140.77	138.98–142.51	↓ 1	Open weights (unrestricted)
122	Gemini 2.5 Flash (Apr 2025)Google DeepMind	140.73	138.34–143.03	↓ 3	API access
123	o3-miniOpenAI	140.65	138.95–142.27	↓ 2	API access
124	gpt-oss-120bOpenAI	140.21	138.29–142.03	↓ 2	Open weights (unrestricted)
125	Grok-3 minixAI	139.92	138.03–141.71	↓ 2	API access
126	Qwen3.5-9BAlibaba	139.35	137.09–141.54	↓ 2	Open weights (restricted use)
127	GPT-5 nanoOpenAI	139.16	137.20–140.99	↓ 2	API access
128	DeepSeek-R1DeepSeek	138.65	136.72–140.44	↓ 2	Open weights (unrestricted)
129	Qwen3-235B-A22BAlibaba	138.42	135.96–140.75	↓ 2	Open weights (unrestricted)
130	Qwen3-30B-A3B-Thinking (Jul 2025)Alibaba	138.40	135.72–140.94	↓ 2	Open weights (unrestricted)
131	Qwen3-235B-A22B-Instruct (Jul 2025)Alibaba	138.11	135.88–140.22	↓ 2	Open weights (unrestricted)
132	Grok 3xAI	137.51	135.63–139.35	↓ 2	API access
133	Qwen3-32BAlibaba	136.97	134.54–139.35	↓ 2	Open weights (unrestricted)
134	QwQ-32BAlibaba	136.53	133.64–139.30	↓ 2	Open weights (unrestricted)
135	GPT-4.5OpenAI	136.48	134.38–138.52	↓ 2	API access
136	Qwen3-14BAlibaba	136.18	133.54–138.68	↓ 2	Open weights (unrestricted)
137	GPT-4.1OpenAI	136.17	134.20–138.04	↓ 2	API access
138	o1-miniOpenAI	135.38	133.16–137.48	↓ 2	API access
139	Gemini 2.0 ProGoogle DeepMind	135.26	132.42–137.94	↓ 2	Hosted access (no API)
140	DeepSeek-V3 (Mar 2025)DeepSeek	135.24	133.00–137.30	↓ 2	Open weights (restricted use)
141	Qwen3-30B-A3B-Instruct (Jul 2025)Alibaba	135.19	132.28–137.99	↓ 2	Open weights (unrestricted)
142	gpt-oss-20bOpenAI	135.04	132.35–137.51	↓ 2	Open weights (unrestricted)
143	GPT-4.1 miniOpenAI	134.27	132.09–136.26	↓ 2	API access
144	Qwen3-30B-A3BAlibaba	134.24	131.32–136.90	↓ 2	Open weights (unrestricted)
145	Gemini 2.0 Flash (Feb 2025)Google DeepMind,Google	134.05	131.74–136.20	↓ 2	API access
146	o1-previewOpenAI	133.87	131.38–136.19	↓ 2	API access
147	Gemini 2.0 Flash Thinking (Jan 2025)Google DeepMind,Google	133.74	130.68–136.54	↓ 2	API access
148	Claude 3.5 Sonnet (October 2024)Anthropic	133.46	131.16–135.67	↓ 2	API access
149	Gemini 2.5 Flash-Lite (Jun 2025)Google DeepMind	133.15	129.39–136.69	↓ 2	API access
150	Mistral Medium 3Mistral AI	132.87	130.14–135.36	↓ 2	API access
151	Qwen3-8BAlibaba	132.21	129.07–135.09	↓ 2	Open weights (unrestricted)
152	Llama 4 MaverickMeta AI	132.13	129.79–134.28	↓ 2	Open weights (restricted use)
153	DeepSeek-V3DeepSeek	131.74	129.23–134.11	↓ 2	Open weights (restricted use)
154	Qwen2.5-MaxAlibaba	131.23	128.58–133.73	↓ 2	API access
155	Gemini 1.5 Pro (Sept 2024)Google DeepMind	130.43	127.72–132.91	↓ 2	API access
156	Claude 3.5 SonnetAnthropic	130.00	Not supplied	↓ 2	API access
157	Grok-2 (Dec 2024)xAI	129.78	127.05–132.33	↓ 2	API access
158	Phi-4Microsoft Research	129.11	126.06–131.88	↓ 2	Open weights (unrestricted)
159	GPT-4o (Nov 2024)OpenAI	128.95	126.22–131.41	↓ 2	API access
160	GPT-4o (May 2024)OpenAI	128.20	125.44–130.87	↓ 2	API access
161	Llama 3.1-405BMeta AI	127.89	125.01–130.59	↓ 2	Open weights (restricted use)
162	Llama 4 ScoutMeta AI	127.74	124.84–130.37	↓ 2	Open weights (restricted use)
163	Gemma 3 27BGoogle DeepMind	127.67	124.63–130.46	↓ 2	Open weights (restricted use)
164	Qwen2.5-72BAlibaba	127.25	124.47–129.84	↓ 2	Open weights (unrestricted)
165	Gemini 1.5 Flash (Sep 2024)Google DeepMind	127.15	124.01–129.96	↓ 2	API access
166	GPT-4o (Aug 2024)OpenAI	127.13	124.09–129.87	↓ 2	API access
167	GPT-4.1 nanoOpenAI	126.89	123.84–129.60	↓ 2	API access
168	Mistral Large 2 (Jul 2024)Mistral AI	126.35	123.10–129.31	↓ 2	Open weights (non-commercial)
169	Llama 3.3 70BMeta AI	126.19	123.22–128.93	↓ 2	Open weights (restricted use)
170	GPT-4 Turbo (Apr 2024)OpenAI	125.87	122.73–128.78	↓ 2	API access
171	Claude 3 OpusAnthropic	125.57	122.61–128.34	↓ 2	API access
172	Gemini 1.5 Pro (May 2024)Google DeepMind	125.41	122.08–128.48	↓ 2	API access
173	Claude 3.5 HaikuAnthropic	124.98	121.68–128.06	↓ 2	API access
174	GPT-4 Turbo (Nov 2023)OpenAI	124.85	121.50–127.84	↓ 2	API access
175	Llama 3.1-70BMeta AI	124.70	121.40–127.62	↓ 2	Open weights (restricted use)
176	GPT-4o miniOpenAI	124.02	120.70–127.05	↓ 2	API access
177	Llama 3.2 90BMeta AI	123.01	119.13–126.46	↓ 2	Open weights (restricted use)
178	Qwen2-72BAlibaba	122.35	118.67–125.65	↓ 2	Open weights (unrestricted)
179	GPT-4 (Mar 2023)OpenAI	121.87	118.06–125.35	↓ 2	API access
180	Llama 3-70BMeta AI	121.34	117.50–124.79	↓ 2	Open weights (restricted use)
181	DeepSeek-V2 (MoE-236B, May 2024)DeepSeek	120.84	114.06–127.52	↓ 2	Open weights (restricted use)
182	Gemini 1.5 Flash (May 2024)Google DeepMind	120.55	116.55–124.15	↓ 2	API access
183	GPT-4 (Jun 2023)OpenAI	120.19	116.45–123.62	↓ 2	API access
184	Mistral LargeMistral AI	119.87	115.75–123.45	↓ 2	API access
185	Claude 3 SonnetAnthropic	119.54	115.36–123.31	↓ 2	API access
186	Mixtral 8x22BMistral AI	118.83	114.52–122.61	↓ 1	Open weights (unrestricted)
187	Qwen2.5-Coder-32BAlibaba	118.77	110.17–127.30	↓ 3	Open weights (unrestricted)
188	Gemma 2 27BGoogle DeepMind	118.67	114.40–122.45	↓ 2	Open weights (restricted use)
189	Claude 3 HaikuAnthropic	117.73	113.50–121.66	↓ 2	API access
190	Claude 2Anthropic	117.18	112.78–121.12	↓ 2	API access
191	phi-3-small 7.4BMicrosoft	116.54	110.70–121.90	↓ 2	Open weights (unrestricted)
192	phi-3-medium 14BMicrosoft	114.52	109.70–118.84	↓ 2	Open weights (unrestricted)
193	Claude InstantAnthropic	114.30	107.64–120.38	↓ 2	API access
194	PaLM 2-L	114.08	100.04–129.36	↓ 2	Unknown
195	Qwen2.5-Coder-14B	114.03	105.51–122.48	↓ 1	Unknown
196	Mistral NeMoMistral AI	114.02	108.85–118.70	↓ 3	Open weights (unrestricted)
197	Mixtral 8x7BMistral AI	113.36	108.48–117.84	↓ 2	Open weights (unrestricted)
198	GPT-3.5 Turbo (Jan 2024)OpenAI	113.22	108.22–117.77	↓ 2	API access
199	GPT-3.5 Turbo (Nov 2023)OpenAI	113.04	107.83–117.83	↓ 2	API access
200	Gemma 2 9BGoogle DeepMind	112.88	107.01–117.96	↓ 2	Open weights (restricted use)
201	Llama 3.1-8BMeta AI	111.24	105.25–116.30	↓ 2	Open weights (restricted use)
202	phi-3-mini 3.8BMicrosoft	110.84	104.35–116.57	↓ 2	Open weights (unrestricted)
203	Stable Beluga 2Stability AI	109.84	102.28–116.62	↓ 2	Open weights (non-commercial)
204	Llama 3-8BMeta AI	109.56	104.13–114.47	↓ 2	Open weights (restricted use)
205	Qwen2.5-Coder (7B)Alibaba	108.03	98.73–116.43	↓ 2	Open weights (unrestricted)
206	Falcon-180BTechnology Innovation Institute	107.40	98.58–115.41	↓ 2	Open weights (restricted use)
207	GPT-3.5 Turbo (Jun 2023)OpenAI	106.42	98.13–113.83	↓ 2	API access
208	Llama 2-70BMeta AI	106.05	100.05–111.50	↓ 2	Open weights (restricted use)
209	Gemma 7BGoogle DeepMind	103.62	95.87–110.27	↓ 2	Open weights (restricted use)
210	Mistral 7B v0.1Mistral AI	103.28	95.65–109.83	↓ 2	Open weights (unrestricted)
211	Yi-9B	102.58	91.48–112.30	↓ 2	Unknown
212	Qwen-14BAlibaba	102.24	93.53–109.75	↓ 2	Open weights (restricted use)
213	LLaMA-65BMeta AI	102.10	93.86–109.45	↓ 2	Open weights (non-commercial)
214	DeepSeek-Coder-V2-Lite-Base	100.07	89.85–109.08	↓ 2	Unknown
215	Falcon 2 11BTechnology Innovation Institute	99.81	88.60–109.45	↓ 2	Open weights (restricted use)
216	Nemotron-4 15BNvidia	99.68	90.36–107.69	↓ 2	Unreleased
217	internlm-20b	99.58	89.64–108.02	↓ 2	Unknown
218	PaLM 2-M	98.66	85.99–110.19	↓ 2	Unknown
219	Qwen2.5-Coder-3B	97.50	84.35–109.24	↓ 2	Unknown
220	LLaMA-33BMeta AI	97.21	87.97–105.03	↓ 2	Open weights (non-commercial)
221	Llama 2-13BMeta AI	97.07	87.98–104.76	↓ 2	Open weights (restricted use)
222	Phi-2Microsoft	96.82	87.36–105.00	↓ 1	Open weights (unrestricted)
223	StarCoder 2 15BHugging Face,ServiceNow,Nvidia,BigCode	96.72	85.47–106.35	↓ 3	Open weights (restricted use)
224	Yi 6B01.AI	94.89	84.82–104.08	↓ 2	Open weights (restricted use)
225	PaLM 2-S	94.63	81.05–106.37	↓ 1	Unknown
226	Llama 2-34BMeta AI	94.57	84.74–103.21	↓ 3	Unreleased
227	Baichuan2-13BBaichuan	92.57	81.75–102.05	↓ 2	Open weights (restricted use)
228	Falcon-40BTechnology Innovation Institute	91.43	80.52–100.53	↓ 2	Open weights (unrestricted)
229	Qwen2.5-Coder (1.5B)Alibaba	90.04	77.34–100.75	↓ 1	Open weights (unrestricted)
230	Qwen-7BAlibaba	89.89	78.55–99.43	↓ 3	Open weights (restricted use)
231	LLaMA-13BMeta AI	86.13	74.18–96.19	↓ 2	Open weights (non-commercial)
232	MPT-30BMosaicML	86.11	73.90–96.08	↓ 2	Open weights (unrestricted)
233	internlm-7b	85.52	73.28–95.86	↓ 2	Unknown
234	Baichuan 2-7BBaichuan	84.45	71.97–95.27	↓ 2	Open weights (restricted use)
235	INTELLECT-1Prime Intellect,Hugging Face,Arcee AI	84.33	71.85–95.29	↓ 2	Unknown
236	Llama 2-7BMeta AI	84.06	71.63–94.62	↓ 2	Open weights (restricted use)
237	LLaMA-7BMeta AI	78.34	64.57–90.16	↓ 2	Open weights (non-commercial)
238	DeepSeek Coder 33BDeepSeek,Peking University	77.36	61.44–90.41	↓ 2	Open weights (restricted use)
239	vicuna-13b-v1.1	77.12	62.34–89.76	↓ 2	Unknown
240	Gemma 2BGoogle DeepMind	76.82	62.52–88.92	↓ 2	Open weights (restricted use)
241	CodeQwen1.5-7B	76.31	60.21–89.98	↓ 2	Unknown
242	chatglm2-6b	74.82	58.68–88.57	↓ 2	Unknown
243	MPT-7BMosaicML	74.62	59.71–87.23	↓ 2	Open weights (unrestricted)
244	Falcon-7BTechnology Innovation Institute	74.29	59.31–86.58	↓ 2	Open weights (unrestricted)
245	StarCoder 2 7BHugging Face,ServiceNow,Nvidia,BigCode	73.70	57.08–87.41	↓ 2	Open weights (restricted use)
246	XGen-7BSalesforce	71.39	55.04–85.22	↓ 2	Open weights (unrestricted)
247	Phi-1.5Microsoft	69.69	51.92–84.34	↓ 2	Open weights (unrestricted)
248	Qwen2.5-Coder-0.5B	69.34	50.58–85.10	↓ 2	Unknown
249	DeepSeek Coder 6.7BDeepSeek,Peking University	68.76	50.56–83.82	↓ 2	Open weights (restricted use)
250	StarCoder 2 3BHugging Face,ServiceNow,Nvidia,BigCode	68.01	49.56–83.31	↓ 2	Open weights (restricted use)
251	open_llama_7b	65.60	47.60–80.66	↓ 2	Unknown
252	Qwen-1_8B	61.07	42.70–77.15	↓ 2	Unknown
253	RedPajama-INCITE-7B-Base	59.10	39.41–76.02	↓ 2	Unknown
254	Dolly 2.0-12bDatabricks	57.08	36.59–73.99	↓ 2	Open weights (unrestricted)
255	Cerebras-GPT-13BCerebras Systems	50.54	27.88–69.77	↓ 2	Open weights (unrestricted)

Ranking: descending published ECI; exact ties share a rank. Changes compare consecutive snapshots only when the source method is unchanged. Source: Epoch AI ↗ · CC BY 4.0 · Ordered and reformatted by RankMondo.

STRUCTURED UPDATE · NO AI INFERENCE

What changed?

253 models changed rank or score; 2 models entered this dataset; 0 left coverage.

Historical snapshots: 3. No earlier history has been reconstructed.

Read the ranking method ↗

AI model rankings.

The capability universe

Capability, with context

What changed?