Who is building notable and frontier AI models, and in which countries?

Counted from Epoch AI’s model catalogue, which has a stated inclusion rule and years of history, not from what our feeds happened to mention. Releases reported in our coverage are listed separately at the foot of the page.

Sources on this page

Sources◆ 2 curated○ 1 coverage3 ▾

United Kingdom · notable models · Q3 2026

of 34 worldwide · no change vs Q2 2026

◆ Epoch AI · as of 25 Oct 2026 · n = 340

United Kingdom · share of notable models · last 4 quarters

0.0%

0 of 122 worldwide · peers: US 60.7% · CN 30.3% · FR 0.0%

◆ Epoch AI · as of 25 Oct 2026 · n = 340

Frontier models · last 12 months

Epoch frontier definition · United Kingdom 0

◆ Epoch AI · as of 25 Oct 2026 · n = 340

Notable models per quarter, by developer country

Notable Frontier only

Focus country and named peers · one column per quarter, Q4 2023 to Q3 2026 · each row on its own scale; the figure on the right is the count · EU-27 shown, never ranked

GBUnited Kingdom

Training compute of frontier models over time

Capability against price

Organisations behind the most notable models

OpenAIUS27
AlibabaCN11
AnthropicUS11
Google DeepMindUS

Latest notable models

Gemini 4 Argon ↗Google DeepMindUSUnknown2026-09-30Google DeepMind · US · Unknown
GPT-6.1 Sol ↗OpenAIUSAPI only2026-09-29OpenAI · US · API only
Claude Sonnet 5.5 ↗AnthropicUSAPI only2026-09-28Anthropic · US · API only

Reported in our coverage — not counted above

Model launches our feeds wrote about in the last 14 days. Rows with no named organisation are left out (1 left out). Each opens its story.

View as table

Notable models per quarter by developer country
Country	Q4 2023	Q1 2024	Q2 2024	Q3 2024	Q4 2024	Q1 2025	Q2 2025	Q3 2025	Q4 2025	Q1 2026	Q2 2026	Q3 2026
United Kingdom	1	1	2	0	0	1	0	0	0	0	0	0
United States	33	12	13	11	21	15	11	14	24	9	20	21
China	13	5	5	11	10	9	6	12	8	10	10	9
France	3	1	0	2	1	1	0	0	0	0	0	0
Germany	0	0	0	1	0	0	0	0	0	0	0	0
Japan	2	0	0	0	0	0	0	0	0	0	0	0
South Korea	2	0	0	0	1	1	0	2	5	0	2	3
EU-27 (aggregate)	4	1	0	3	1	1	0	0	0	0	0	0
Rest of world	1	1	1	3	0	0	0	1	0	0	0	0
Worldwide (distinct models)	50	19	19	28	32	24	17	29	37	19	32	34

View as table

Training compute of frontier models
Model	Developer	Published	log10 FLOP	Countries
GPT-6 Astra	OpenAI	3 Sept 2026	26.97	US
Grok 4	xAI	9 Jul 2025	26.70	US
Llama 4 Behemoth (preview)	Meta AI	5 Apr 2025	25.71	US
Llama Nemotron Ultra 253B	Nvidia	18 Mar 2025	25.59	US
GPT-4.5	OpenAI	27 Feb 2025	26.58	US
Grok 3	xAI	17 Feb 2025	26.54	US
Grok-2	xAI	13 Aug 2024	25.47	US
Llama 3.1-405B	Meta AI	23 Jul 2024	25.58	US
Claude 3.5 Sonnet	Anthropic	20 Jun 2024	25.43	US
Nemotron-4 340B	Nvidia	14 Jun 2024	25.26	US
Gemini 1.0 Ultra	Google DeepMind	6 Dec 2023	25.70	US
Inflection-2	Inflection AI	22 Nov 2023	25.00	US
Amazon Titan	Amazon	28 Sept 2023	24.68	US
Falcon-180B	Technology Innovation Institute	6 Sept 2023	24.58	AE
Claude 2	Anthropic	11 Jul 2023	24.59	US
GPT-4 (Jun 2023)	OpenAI	13 Jun 2023	25.32	US
PaLM 2	Google	10 May 2023	24.87	US
SenseChat	SenseTime	10 Apr 2023	24.59	HK
GPT-4 (Mar 2023)	OpenAI	15 Mar 2023	25.32	US
Flan-PaLM 540B	Google	20 Oct 2022	24.40	US
U-PaLM (540B)	Google	20 Oct 2022	24.40	US
Minerva (540B)	Google	29 Jun 2022	24.44	US
PaLM (540B)	Google Research	4 Apr 2022	24.40	US
GPT-3.5 (davinci-002)	OpenAI	15 Mar 2022	24.41	US
ERNIE 3.0 Titan	Baidu,Peng Cheng Laboratory	23 Dec 2021	24.02	CN
EXAONE 1.0	LG	14 Dec 2021	24.23	KR
Gopher (280B)	DeepMind	8 Dec 2021	23.80	GB
Yuan 1.0	Inspur	12 Oct 2021	23.55	CN
Megatron-Turing NLG 530B	Microsoft,Nvidia	11 Oct 2021	23.93	US
HyperCLOVA 204B	NAVER	10 Sept 2021	23.30	KR
FLAN 137B	Google Research	3 Sept 2021	24.31	US
Jurassic-1-Jumbo	AI21 Labs	11 Aug 2021	23.57	IL
Wu Dao 2.0	Beijing Academy of Artificial Intelligence / BAAI	31 May 2021	24.19	CN
Switch	Google	11 Jan 2021	22.91	US
mT5-XXL	Google,Google Research	20 Oct 2020	22.91	US
GShard (dense)	Google	30 Jun 2020	22.68	US
iGPT-XL	OpenAI	17 Jun 2020	22.52	US
GPT-3 175B (davinci)	OpenAI	28 May 2020	23.50	US
Meena	Google Brain	28 Jan 2020	23.05	US
OpenAI Five	OpenAI	13 Dec 2019	22.83	US
Noisy Student (L2)	Carnegie Mellon University (CMU),Google	11 Nov 2019	22.42	US
XLM-RoBERTa	Facebook AI	5 Nov 2019	22.32	US
AlphaStar	DeepMind	30 Oct 2019	23.03	GB
T5-11B	Google	23 Oct 2019	22.52	US
T5-3B	Google	23 Oct 2019	21.95	US
Megatron-BERT	Nvidia	17 Sept 2019	22.34	US
Megatron-LM (8.3B)	Nvidia	17 Sept 2019	21.96	US
RoBERTa Large	Facebook,University of Washington	1 Jul 2019	21.93	US
XLNet	Carnegie Mellon University (CMU),Google Brain	1 Jun 2019	21.79	US
Grover-Mega	University of Washington	29 May 2019	21.67	US
GPT-2 (355M)	OpenAI	14 Feb 2019	21.36	US
GPT-2 (774M)	OpenAI	14 Feb 2019	21.69	US
BigGAN-deep 512x512	Heriot-Watt University,DeepMind	28 Sept 2018	21.26	GB
ResNeXt-101 32x48d	Facebook	2 May 2018	21.94	US

How notable models can be accessed

Open weights, unrestricted3232 · 26.2%
Open weights, restricted1212 · 9.8%
API access6868 · 55.7%
Hosted access00 · 0%
Unreleased77 · 5.7%
Unknown33 · 2.5%

2025-Q4..2026-Q3 · 122 notable models · 3 with no recorded access terms

View as table

Access of notable models
Access	Models	Share
Open weights, unrestricted	32	26.2%
Open weights, restricted	12	9.8%
API access	68	55.7%
Hosted access	0	0%
Unreleased	7	5.7%
Unknown	3	2.5%

More on Open vs Closed ▸

Universe Epoch ‘notable’ models (its inclusion rule); country = developer organisation's country; a joint release counts once per countryn 122 modelsPeriod Q4 2025 – Q3 2026 · as of 25 Oct 2026Gap Models whose access Epoch does not record are shown as unknown, not droppedGap 3 with an access value we do not map: Limited accessData: Epoch AI, ‘Data on AI Models’, CC BY 4.0, epoch.ai/data/ai-models

View as table

Capability index and price
Model	Index	US$ per million tokens	Weights
Claude Opus 5.5 (Max, Default Fallback)	57.6	8	Unknown
Claude Fable 5.1 (Max, Default Fallback)	53.4	20	Unknown
GPT-6 Astra (Max)	52.7	20	Unknown
GPT-6 Astra (Xhigh)	52.4	20	Unknown
GPT-6.1 Sol (Xhigh)	51	4	Unknown
GPT-6.1 Sol (High)	50.2	4	Unknown
Claude Opus 5 (Xhigh)	49.7	10	Unknown
Claude Fable 5.1 (Medium, Default Fallback)	48.9	20	Unknown
Claude Fable 5.1 (Low, Default Fallback)	46.8	20	Unknown
Claude Sonnet 5.5 (High, Default Fallback)	46.8	4	Unknown
Grok 4.7 (High)	46.3	3	Unknown
MiMo-V2.6-Pro	46.3	0.54375	Unknown
Step 5 Preview	43.7	1.425	Unknown
Grok 4.6 (Medium)	42.8	3	Unknown
Grok 4.7 (Low)	42.2	3	Unknown
GLM-5.3-Flash	41.8	0.2375	Unknown
Claude Haiku 5.5 (Xhigh)	41.2	0.2	Unknown
Claude Sonnet 5.5 (Medium, Default Fallback)	40.8	4	Unknown
Muse Spark 1.2 (Xhigh)	39.6	2	Unknown
Claude Opus 5 (Low)	39.4	10	Unknown
Muse Spark Open (Xhigh)	38.9	2	Unknown
GPT-5.5 (Xhigh)	38.4	11.25	Unknown
Mistral Large 4 Preview	38.4	2.065	Unknown
Claude Sonnet 5 (Max)	38.2	4	Unknown
MiMo-V2.6-Flash	37.9	0.17500000000000002	Unknown
Claude Haiku 5.5 (High)	37.8	0.2	Unknown
GPT-5.6 Luna (Max)	37.3	0.45	Unknown
GPT-6 Luna (Xhigh)	34.6	0.2	Unknown
GPT-6 Sol (Low)	34.2	4	Unknown
GPT-5.6 Sol (Low)	33.5	8	Unknown
Gemini 3.5 Flash (High)	32.6	3.375	Unknown
Claude Opus 4.7 (Non-reasoning, High)	30.9	10	Unknown
GPT-5.6 Terra (Medium)	30.1	4.5	Unknown
GPT-6 Luna (Medium)	29.9	0.2	Unknown
Claude Haiku 5.5 (Low)	29.5	0.2	Unknown
MiniMax-M3	29.2	0.5249999999999999	Unknown
Claude Opus 4.5 (Reasoning)	29.1	10	Unknown
GPT-5.2 Codex (Xhigh)	28.5	4.8125	Unknown
Nex-N2-Pro (Based on Qwen3.5-397B-A17B)	28.2	—	Unknown
GLM-5 (Reasoning)	27.9	1.55	Unknown
Grok Build 0.1 0616	27.2	1.25	Unknown
Kimi K2.6 (Reasoning)	27	1.7125	Unknown
GLM-5-Turbo	26.6	—	Unknown
Solar Pro 4	26.6	0.5249999999999999	Unknown
Apodex 1.1	26.4	0.975	Unknown
Claude Opus 4.6 (Non-reasoning, High)	26.4	10	Unknown
MiMo-V2.5-Pro (Reasoning)	26	0.564375	Unknown
Grok 4.20 0309 (Reasoning)	25.2	3	Unknown
MiMo-V2-Omni-0327	25.1	—	Unknown
Inkling (Xhigh)	25	1.7625	Unknown
Claude Sonnet 4.6 (Non-reasoning, High)	24.7	6	Unknown
Ling-3.0-flash-VL	24.6	—	Unknown
Claude Sonnet 5 (Low)	24.3	4	Unknown
Gemini 3.5 Flash (Minimal)	23.8	3.375	Unknown
Claude Opus 4.5 (Non-reasoning)	23.7	10	Unknown
GPT-5.1 Codex (High)	23.7	3.4375	Unknown
Kimi K2.6 (Non-reasoning)	23.6	1.7125	Unknown
GLM 5V Turbo (Reasoning)	23.5	—	Unknown
Qwen3.5 27B (Reasoning)	22.9	0.825	Unknown
MiniMax-M2.5	22.8	0.5249999999999999	Unknown
MiMo-V2-Flash (Feb 2026)	22.4	—	Unknown
Gemini 3.5 Flash-Lite	22.2	0.85	Unknown
MiMo-V2.5	21.8	0.16625	Unknown
Qwen3.5 397B A17B (Non-reasoning)	21.4	1.35	Unknown
GPT-5.6 Luna (Low)	21	0.45	Unknown
MiniMax-M2.1	20.9	0.5249999999999999	Unknown
K2 Horizon 7B	20.6	—	Unknown
Grok 4.1 Fast (Reasoning)	20.4	—	Unknown
Qwen3.5 Omni Plus	20.4	1.5	Unknown
Qwen3.8 27B (Non-reasoning)	20.2	1.125	Unknown
o3	20.2	3.5	Unknown
Qwen3.6 27B (Non-reasoning)	19.8	1.35	Unknown
GPT-5.4 mini (Medium)	19.7	1.6875	Unknown
K-EXAONE 2.0 0803	19.7	—	Unknown
JT-35B-Flash	18.7	—	Unknown
Qwen3.5 397B A17B (Reasoning)	18.4	1.35	Unknown
Claude 3.7 Sonnet (Reasoning)	17.7	—	Unknown
Qwen3.5 122B A10B (Non-reasoning)	17.7	1.1	Unknown
Ling-2.6-1T	17	0.85	Unknown
Step 3.5 Flash 2603	17	0.15000000000000002	Unknown
GPT-5 mini (High)	16.8	0.6875	Unknown
Gemini 2.5 Pro	16.1	3.4375	Unknown
K2 Horizon 3.7B	15.6	—	Unknown
Qwen3.5 122B A10B (Reasoning)	15.6	1.1	Unknown
GLM-4.7-Flash (Reasoning)	14.9	0.15250000000000002	Unknown
DeepSeek V3.1 Terminus (Reasoning)	14.8	1.91375	Unknown
Gemini 2.5 Pro Preview (May' 25)	14.5	3.4375	Unknown
Nova 2.0 Pro Preview (Medium)	14.2	3.4375	Unknown
DeepSeek V3.2 Exp (Non-reasoning)	13.9	0.315	Unknown
Nova 2.0 Omni (Medium)	13.6	—	Unknown
DeepSeek V3.1 (Reasoning)	13.5	0.855	Unknown
Qwen3.5 9B (Non-reasoning)	13.3	0.19	Unknown
Command A+	13.1	0	Unknown
Gemini 2.5 Flash (Reasoning)	13.1	0.85	Unknown
Nemotron 3.5 Lightning	12.9	0.095	Unknown
GLM-4.5 (Reasoning)	12.8	—	Unknown
GPT-4.1	12.7	3.5	Unknown
Granite 4.2 30B	12.7	0.2825	Unknown
o3-mini	12.5	1.9250000000000003	Unknown
JT-MINI	12.2	—	Unknown
Grok 3	12.1	6	Unknown
Qwen3 235B A22B 2507 Instruct	12	0.3675	Unknown
Qwen3 Coder 480B A35B Instruct	11.9	0.63125	Unknown
Sonar Reasoning Pro	11.8	—	Unknown
HyperNova 60B 2605 (High, Based on gpt-oss-120b)	11.7	—	Unknown
DeepSeek R1 (Jan '25)	11.4	2.5	Unknown
HyperCLOVA X SEED Think (32B)	11.4	—	Unknown
o1-preview	11.4	28.875	Unknown
Mistral Small 4 (Reasoning)	11.3	0.26249999999999996	Unknown
Qwen3.5 9B (Reasoning)	11.2	0.15125	Unknown
Grok 4 Fast (Non-reasoning)	11.1	0.275	Unknown
o3-mini (High)	11	1.9250000000000003	Unknown
G9v3-3B	10.8	—	Unknown
Mercury 2	10.8	0.375	Unknown
Qwen3.5 4B (Non-reasoning)	10.8	0.06	Unknown
INTELLECT-3 (Based on GLM-4.5-Air)	10.6	—	Unknown
MiniMax M1 40k	10	—	Unknown
gpt-oss-20b (Low)	10	0.1275	Unknown
North Mini Code	9.9	0	Unknown
Qwen3 Next 80B A3B Instruct	9.6	0.35000000000000003	Unknown
Gemini 2.0 Flash Thinking Experimental (Jan '25)	9.4	—	Unknown
Gemma 4 12B (Non-reasoning)	9.4	0.15000000000000002	Unknown
Ling-1T	9.2	—	Unknown
Mistral Medium 3.1	9.2	—	Unknown
Granite 4.2 3B	9.1	0.0525	Unknown
Mistral Small 4 (Non-reasoning)	9	0.26249999999999996	Unknown
gpt-oss-20b (High)	9	0.0925	Unknown
Claude 3.5 Haiku	8.9	—	Unknown
Llama 3.3 Nemotron Super 49B v1 (Reasoning)	8.9	—	Unknown
Qwen3 4B 2507 (Reasoning)	8.8	—	Unknown
Sonar Reasoning	8.7	—	Unknown
Devstral 2	8.6	—	Unknown
DeepSeek V3 (Dec '24)	8.5	0.4625	Unknown
GLM-4.6V (Non-reasoning)	8.4	0.44999999999999996	Unknown
Qwen3 235B A22B (Non-reasoning)	8.3	—	Unknown
Magistral Small 1	8.2	—	Unknown
Nova 2.0 Omni (Non-reasoning)	8.2	—	Unknown
Claude 3.5 Sonnet (Oct '24)	7.9	6	Unknown
Gemini 1.5 Pro (Sep '24)	7.9	—	Unknown
Solar Pro 2 (Preview) (Non-reasoning)	7.9	—	Unknown
GLM-4.5V (Reasoning)	7.6	0.8999999999999999	Unknown
Qwen3 30B A3B (Reasoning)	7.6	0.23875000000000002	Unknown
Sonar Pro	7.6	—	Unknown
Llama Nemotron Super 49B v1.5 (Non-reasoning)	7.4	—	Unknown
Gemini 2.0 Flash-Lite (Preview)	7.3	—	Unknown
Qwen3 VL 8B Instruct	7.3	—	Unknown
Qwen3 4B (Reasoning)	7.2	—	Unknown
GPT-5 nano (Minimal)	7.1	0.1375	Unknown
Gemini 1.5 Flash (Sep '24)	7.1	—	Unknown
Grok 2 (Dec '24)	7.1	—	Unknown
Mistral Small 3.1	7.1	0.09375	Unknown
Olmo 3.1 32B Think	7.1	—	Unknown
Command A	7	4.375	Unknown
Mistral Small 3.2	7	0.10625	Unknown
Qwen2.5 Instruct 32B	6.9	—	Unknown
Mistral Large 2 (Jul '24)	6.8	3	Unknown
NVIDIA Nemotron 3 Nano 30B A3B (Non-reasoning)	6.8	0.08750000000000001	Unknown
NVIDIA Nemotron Nano 9B V2 (Non-reasoning)	6.8	0.08625000000000001	Unknown
GLM-4.5V (Non-reasoning)	6.7	0.8999999999999999	Unknown
Gemini 2.5 Flash-Lite (Non-reasoning)	6.7	0.17500000000000002	Unknown
Hermes 4 - Llama-3.1 70B (Non-reasoning)	6.7	—	Unknown
Qwen2.5 Coder Instruct 32B	6.7	—	Unknown
DeepSeek R1 Distill Llama 8B	6.5	—	Unknown
Gemini 1.5 Pro (May '24)	6.4	—	Unknown
Qwen2.5 Turbo	6.4	—	Unknown
EXAONE 4.0 32B (Non-reasoning)	6.3	—	Unknown
Qwen3.5 2B (Non-reasoning)	6.2	—	Unknown
Hermes 3 - Llama-3.1 70B	6	0.7	Unknown
Qwen3 Omni 30B A3B Instruct	6	0.43	Unknown
Nova Micro	5.9	0.061250000000000006	Unknown
Gemini 1.0 Ultra	5.8	—	Unknown
Phi-4 Multimodal Instruct	5.8	0.14	Unknown
MiniCPM-V 4.6 1.3B	5.7	—	Unknown
Claude 3 Haiku	5.6	0.5	Unknown
Olmo 3 7B Think	5.6	—	Unknown
DeepSeek-V2-Chat	5.5	—	Unknown
GPT-3.5 Turbo	5.5	0.75	Unknown
Ling-mini-2.0	5.5	—	Unknown
Mistral Medium	5.5	—	Unknown
Mistral Small (Feb '24)	5.5	—	Unknown
LFM 40B	5.4	—	Unknown
Qwen Chat 72B	5.4	—	Unknown
Qwen3.5 0.8B (Non-reasoning)	5.4	—	Unknown
DBRX Instruct	5.3	—	Unknown
Sarvam M (Reasoning, Based on Mistral Small 3.1)	5.3	0	Unknown
Gemma 3 270M	5.1	—	Unknown
Granite 4.0 Micro	5.1	—	Unknown
Mixtral 8x7B Instruct	5.1	0.5125	Unknown
Mistral 7B Instruct	5	0.16249999999999998	Unknown
Molmo2-8B	5	—	Unknown
Gemma 3 27B Instruct	4.9	0	Unknown
Granite 3.3 8B (Non-reasoning)	4.9	0.08499999999999999	Unknown
Qwen3 1.7B (Non-reasoning)	4.9	—	Unknown
Gemma 3 4B Instruct	4.8	0	Unknown
LFM2 1.2B	4.8	—	Unknown
LFM2.5-VL-1.6B	4.8	—	Unknown
Qwen3 0.6B (Non-reasoning)	4.8	—	Unknown
Gemma 3 12B Instruct	3.8	0	Unknown

View as table

Organisations with the most notable models
Organisation	Countries	Notable models
OpenAI	US	27
Alibaba	CN	11
Anthropic	US	11
Google DeepMind	US	11
Z.ai (Zhipu AI)	CN	5
DeepSeek	CN	4
Meta AI	US	4
Moonshot	CN	4
xAI	US	4
Character.ai	US	3