stair-lab/platinum_detect · Datasets at Hugging Face

name stringlengths 10 47	display_name stringlengths 5 41	short_display_name stringlengths 5 41 ⌀	description stringlengths 3 534	creator_organization stringlengths 4 22	access stringclasses 3 values	todo bool 2 classes	release_date stringdate 2019-10-23 00:00:00 2024-12-11 00:00:00 ⌀	num_parameters float64 1 1,210B ⌀
01-ai/yi-34b	Yi (34B)	null	The Yi models are large language models trained from scratch by developers at 01.AI.	01.AI	open	false	2023/11/2	34,000,000,000
01-ai/yi-34b-chat	Yi Chat (34B)	Yi Chat (34B)	The Yi models are large language models trained from scratch by developers at 01.AI.	01.AI	open	false	2023/11/23	34,000,000,000
01-ai/yi-6b	Yi (6B)	null	The Yi models are large language models trained from scratch by developers at 01.AI.	01.AI	open	false	2023/11/2	6,000,000,000
01-ai/yi-large-preview	Yi Large (Preview)	Yi Large (Preview)	The Yi models are large language models trained from scratch by developers at 01.AI. ([tweet](https://x.com/01AI_Yi/status/1789894091620458667))	01.AI	limited	false	2024/5/12	null
ai21/j1-grande	J1-Grande v1 (17B)	null	Jurassic-1 Grande (17B parameters) with a "few tweaks" to the training process ([docs](https://studio.ai21.com/docs/jurassic1-language-models/), [tech report](https://uploads-ssl.webflow.com/60fd4503684b466578c0d307/61138924626a6981ee09caf6_jurassic_tech_paper.pdf)).	AI21 Labs	limited	false	2022/5/3	17,000,000,000
ai21/j1-grande-v2-beta	J1-Grande v2 beta (17B)	null	Jurassic-1 Grande v2 beta (17B parameters)	AI21 Labs	limited	false	2022/10/28	17,000,000,000
ai21/j1-jumbo	J1-Jumbo v1 (178B)	null	Jurassic-1 Jumbo (178B parameters) ([docs](https://studio.ai21.com/docs/jurassic1-language-models/), [tech report](https://uploads-ssl.webflow.com/60fd4503684b466578c0d307/61138924626a6981ee09caf6_jurassic_tech_paper.pdf)).	AI21 Labs	limited	false	2021/8/11	178,000,000,000
ai21/j1-large	J1-Large v1 (7.5B)	null	Jurassic-1 Large (7.5B parameters) ([docs](https://studio.ai21.com/docs/jurassic1-language-models/), [tech report](https://uploads-ssl.webflow.com/60fd4503684b466578c0d307/61138924626a6981ee09caf6_jurassic_tech_paper.pdf)).	AI21 Labs	limited	false	2021/8/11	7,500,000,000
ai21/j2-grande	Jurassic-2 Grande (17B)	null	Jurassic-2 Grande (17B parameters) ([docs](https://www.ai21.com/blog/introducing-j2))	AI21 Labs	limited	false	2023/3/9	17,000,000,000
ai21/j2-jumbo	Jurassic-2 Jumbo (178B)	null	Jurassic-2 Jumbo (178B parameters) ([docs](https://www.ai21.com/blog/introducing-j2))	AI21 Labs	limited	false	2023/3/9	178,000,000,000
ai21/j2-large	Jurassic-2 Large (7.5B)	null	Jurassic-2 Large (7.5B parameters) ([docs](https://www.ai21.com/blog/introducing-j2))	AI21 Labs	limited	false	2023/3/9	7,500,000,000
ai21/jamba-1.5-large	Jamba 1.5 Large	Jamba 1.5 Large	Jamba 1.5 Large is a long-context, hybrid SSM-Transformer instruction following foundation model that is optimized for function calling, structured output, and grounded generation. ([blog](https://www.ai21.com/blog/announcing-jamba-model-family))	AI21 Labs	open	false	2024/8/22	399,000,000,000
ai21/jamba-1.5-mini	Jamba 1.5 Mini	Jamba 1.5 Mini	Jamba 1.5 Mini is a long-context, hybrid SSM-Transformer instruction following foundation model that is optimized for function calling, structured output, and grounded generation. ([blog](https://www.ai21.com/blog/announcing-jamba-model-family))	AI21 Labs	open	false	2024/8/22	51,600,000,000
ai21/jamba-instruct	Jamba Instruct	Jamba Instruct	Jamba Instruct is an instruction tuned version of Jamba, which uses a hybrid Transformer-Mamba mixture-of-experts (MoE) architecture that interleaves blocks of Transformer and Mamba layers. ([blog](https://www.ai21.com/blog/announcing-jamba-instruct))	AI21 Labs	limited	false	2024/5/2	52,000,000,000
aisingapore/llama3-8b-cpt-sea-lionv2-base	Llama 3 CPT SEA-Lion v2 (8B)	Llama 3 CPT SEA-Lion v2 (8B)	Llama 3 CPT SEA-Lion v2 (8B) is a multilingual model which was continued pre-trained on 48B additional tokens, including tokens in Southeast Asian languages.	AI Singapore	open	false	2024/7/31	80,300,000,000
aisingapore/llama3-8b-cpt-sea-lionv2.1-instruct	Llama 3 CPT SEA-Lion v2.1 Instruct (8B)	Llama 3 CPT SEA-Lion v2.1 Instruct (8B)	Llama 3 CPT SEA-Lion v2.1 Instruct (8B) is a multilingual model which has been fine-tuned with around 100,000 English instruction-completion pairs alongside a smaller pool of around 50,000 instruction-completion pairs from other Southeast Asian languages, such as Indonesian, Thai and Vietnamese.	AI Singapore	open	false	2024/8/21	80,300,000,000
aisingapore/sea-lion-7b-instruct	SEA-LION Instruct (7B)	SEA-LION Instruct (7B)	SEA-LION is a collection of language models which has been pretrained and instruct-tuned on languages from the Southeast Asia region. It utilizes the MPT architecture and a custom SEABPETokenizer for tokenization.	AI Singapore	open	false	2023/2/24	7,000,000,000
AlephAlpha/luminous-base	Luminous Base (13B)	null	Luminous Base (13B parameters) ([docs](https://docs.aleph-alpha.com/docs/introduction/luminous/))	Aleph Alpha	limited	false	2022/1/1	13,000,000,000
AlephAlpha/luminous-extended	Luminous Extended (30B)	null	Luminous Extended (30B parameters) ([docs](https://docs.aleph-alpha.com/docs/introduction/luminous/))	Aleph Alpha	limited	false	2022/1/1	30,000,000,000
AlephAlpha/luminous-supreme	Luminous Supreme (70B)	null	Luminous Supreme (70B parameters) ([docs](https://docs.aleph-alpha.com/docs/introduction/luminous/))	Aleph Alpha	limited	false	2022/1/1	70,000,000,000
allenai/olmo-1.7-7b	OLMo 1.7 (7B)	OLMo 1.7 (7B)	OLMo is a series of Open Language Models trained on the Dolma dataset. The instruct versions was trained on the Tulu SFT mixture and a cleaned version of the UltraFeedback dataset.	Allen Institute for AI	open	false	2024/4/17	7,000,000,000
allenai/olmo-7b	OLMo (7B)	OLMo (7B)	OLMo is a series of Open Language Models trained on the Dolma dataset.	Allen Institute for AI	open	false	2024/2/1	7,000,000,000
anthropic/claude-2.0	Anthropic Claude 2.0	null	Claude 2.0 is a general purpose large language model developed by Anthropic. It uses a transformer architecture and is trained via unsupervised learning, RLHF, and Constitutional AI (including both a supervised and Reinforcement Learning (RL) phase). ([model card](https://efficient-manatee.files.svdcdn.com/production/images/Model-Card-Claude-2.pdf))	Anthropic	limited	false	2023/7/11	null
anthropic/claude-2.1	Anthropic Claude 2.1	null	Claude 2.1 is a general purpose large language model developed by Anthropic. It uses a transformer architecture and is trained via unsupervised learning, RLHF, and Constitutional AI (including both a supervised and Reinforcement Learning (RL) phase). ([model card](https://efficient-manatee.files.svdcdn.com/production/images/Model-Card-Claude-2.pdf))	Anthropic	limited	false	2023/11/21	null
anthropic/claude-3-5-haiku-20241022	Claude 3.5 Haiku (20241022)	Claude 3.5 Haiku (20241022)	Claude 3.5 Haiku is a Claude 3 family model which matches the performance of Claude 3 Opus at a similar speed to the previous generation of Haiku ([blog](https://www.anthropic.com/news/3-5-models-and-computer-use)).	Anthropic	limited	false	2024/11/4	null
anthropic/claude-3-5-sonnet-20240620	Claude 3.5 Sonnet (20240620)	Claude 3.5 Sonnet (20240620)	Claude 3.5 Sonnet is a Claude 3 family model which outperforms Claude 3 Opus while operating faster and at a lower cost. ([blog](https://www.anthropic.com/news/claude-3-5-sonnet))	Anthropic	limited	false	2024/6/20	null
anthropic/claude-3-5-sonnet-20241022	Claude 3.5 Sonnet (20241022)	Claude 3.5 Sonnet (20241022)	Claude 3.5 Sonnet is a Claude 3 family model which outperforms Claude 3 Opus while operating faster and at a lower cost ([blog](https://www.anthropic.com/news/claude-3-5-sonnet)). This is an upgraded snapshot released on 2024-10-22 ([blog](https://www.anthropic.com/news/3-5-models-and-computer-use)).	Anthropic	limited	false	2024/10/22	null
anthropic/claude-3-haiku-20240307	Claude 3 Haiku (20240307)	Claude 3 Haiku (20240307)	Claude 3 is a a family of models that possess vision and multilingual capabilities. They were trained with various methods such as unsupervised learning and Constitutional AI ([blog](https://www.anthropic.com/news/claude-3-family)).	Anthropic	limited	false	2024/3/13	null
anthropic/claude-3-opus-20240229	Claude 3 Opus (20240229)	Claude 3 Opus (20240229)	Claude 3 is a a family of models that possess vision and multilingual capabilities. They were trained with various methods such as unsupervised learning and Constitutional AI ([blog](https://www.anthropic.com/news/claude-3-family)).	Anthropic	limited	false	2024/3/4	null
anthropic/claude-3-sonnet-20240229	Claude 3 Sonnet (20240229)	Claude 3 Sonnet (20240229)	Claude 3 is a a family of models that possess vision and multilingual capabilities. They were trained with various methods such as unsupervised learning and Constitutional AI ([blog](https://www.anthropic.com/news/claude-3-family)).	Anthropic	limited	false	2024/3/4	null
anthropic/claude-instant-1.2	Anthropic Claude Instant 1.2	null	A lightweight version of Claude, a model trained using reinforcement learning from human feedback ([docs](https://www.anthropic.com/index/introducing-claude)).	Anthropic	limited	false	2023/8/9	null
anthropic/claude-instant-v1	Anthropic Claude Instant V1	null	A lightweight version of Claude, a model trained using reinforcement learning from human feedback ([docs](https://www.anthropic.com/index/introducing-claude)).	Anthropic	limited	false	2023/3/17	null
anthropic/claude-v1.3	Anthropic Claude v1.3	null	A model trained using reinforcement learning from human feedback ([docs](https://www.anthropic.com/index/introducing-claude)).	Anthropic	limited	false	2023/3/17	null
anthropic/stanford-online-all-v4-s3	Anthropic-LM v4-s3 (52B)	null	A 52B parameter language model, trained using reinforcement learning from human feedback [paper](https://arxiv.org/pdf/2204.05862.pdf).	Anthropic	closed	false	2021/12/1	52,000,000,000
Austism/chronos-hermes-13b	Chronos Hermes 13B	Chronos Hermes 13B	Chronos Hermes 13B is a large language model trained on 13 billion parameters. ([blog](https://chronos.ai/chronos-hermes-13b/))	Chronos	open	false	2024/4/18	13,000,000,000
codellama/CodeLlama-13b-Instruct-hf	CodeLlama 13B Instruct	CodeLlama 13B Instruct	CodeLlama 13B Instruct is a large language model trained on 13 billion parameters. ([blog](https://codellama.com/codellama-13b-instruct/))	CodeLlama	open	false	2024/4/18	13,000,000,000
codellama/CodeLlama-34b-Instruct-hf	CodeLlama 34B Instruct	CodeLlama 34B Instruct	CodeLlama 34B Instruct is a large language model trained on 34 billion parameters. ([blog](https://codellama.com/codellama-34b-instruct/))	CodeLlama	open	false	2024/4/18	34,000,000,000
codellama/CodeLlama-70b-Instruct-hf	CodeLlama 70B Instruct	CodeLlama 70B Instruct	CodeLlama 70B Instruct is a large language model trained on 70 billion parameters. ([blog](https://codellama.com/codellama-70b-instruct/))	CodeLlama	open	false	2024/4/18	70,000,000,000
codellama/CodeLlama-7b-Instruct-hf	CodeLlama 7B Instruct	CodeLlama 7B Instruct	CodeLlama 7B Instruct is a large language model trained on 7 billion parameters. ([blog](https://codellama.com/codellama-7b-instruct/))	CodeLlama	open	false	2024/4/18	7,000,000,000
cognitivecomputations/dolphin-2.5-mixtral-8x7b	Dolphin 2.5 Mixtral 8x7B	Dolphin 2.5 Mixtral 8x7B	Dolphin 2.5 Mixtral 8x7B is a multimodal model trained on 8x7B parameters with a 32K token sequence length. ([blog](https://cognitivecomputations.com/dolphin-2.5-mixtral-8x7b/))	Cognitive Computations	open	false	2024/4/18	46,700,000,000
cohere/command	Cohere Command	null	Command is Cohere’s flagship text generation model. It is trained to follow user commands and to be instantly useful in practical business applications. [docs](https://docs.cohere.com/reference/generate) and [changelog](https://docs.cohere.com/changelog)	Cohere	limited	false	2023/9/29	null
cohere/command-light	Cohere Command Light	null	Command is Cohere’s flagship text generation model. It is trained to follow user commands and to be instantly useful in practical business applications. [docs](https://docs.cohere.com/reference/generate) and [changelog](https://docs.cohere.com/changelog)	Cohere	limited	false	2023/9/29	null
cohere/command-medium-beta	Cohere Command beta (6.1B)	null	Cohere Command beta (6.1B parameters) is fine-tuned from the medium model to respond well with instruction-like prompts ([details](https://docs.cohere.ai/docs/command-beta)).	Cohere	limited	false	2022/11/8	6,100,000,000
cohere/command-r	Command R	Command R	Command R is a multilingual 35B parameter model with a context length of 128K that has been trained with conversational tool use capabilities.	Cohere	open	false	2024/3/11	35,000,000,000
cohere/command-r-plus	Command R Plus	Command R Plus	Command R+ is a multilingual 104B parameter model with a context length of 128K that has been trained with conversational tool use capabilities.	Cohere	open	false	2024/4/4	104,000,000,000
cohere/command-xlarge-beta	Cohere Command beta (52.4B)	null	Cohere Command beta (52.4B parameters) is fine-tuned from the XL model to respond well with instruction-like prompts ([details](https://docs.cohere.ai/docs/command-beta)).	Cohere	limited	false	2022/11/8	52,400,000,000
cohere/large-20220720	Cohere large v20220720 (13.1B)	null	Cohere large v20220720 (13.1B parameters), which is deprecated by Cohere as of December 2, 2022.	Cohere	limited	false	2022/7/20	13,100,000,000
cohere/medium-20220720	Cohere medium v20220720 (6.1B)	null	Cohere medium v20220720 (6.1B parameters)	Cohere	limited	false	2022/7/20	6,100,000,000
cohere/medium-20221108	Cohere medium v20221108 (6.1B)	null	Cohere medium v20221108 (6.1B parameters)	Cohere	limited	false	2022/11/8	6,100,000,000
cohere/small-20220720	Cohere small v20220720 (410M)	null	Cohere small v20220720 (410M parameters), which is deprecated by Cohere as of December 2, 2022.	Cohere	limited	false	2022/7/20	410,000,000
cohere/xlarge-20220609	Cohere xlarge v20220609 (52.4B)	null	Cohere xlarge v20220609 (52.4B parameters)	Cohere	limited	false	2022/6/9	52,400,000,000
cohere/xlarge-20221108	Cohere xlarge v20221108 (52.4B)	null	Cohere xlarge v20221108 (52.4B parameters)	Cohere	limited	false	2022/11/8	52,400,000,000
damo/seallm-7b-v2	SeaLLM v2 (7B)	SeaLLM v2 (7B)	SeaLLM v2 is a multilingual LLM for Southeast Asian (SEA) languages trained from Mistral (7B). ([website](https://damo-nlp-sg.github.io/SeaLLMs/))	Alibaba DAMO Academy	open	false	2024/2/2	7,000,000,000
damo/seallm-7b-v2.5	SeaLLM v2.5 (7B)	SeaLLM v2.5 (7B)	SeaLLM is a multilingual LLM for Southeast Asian (SEA) languages trained from Gemma (7B). ([website](https://damo-nlp-sg.github.io/SeaLLMs/))	Alibaba DAMO Academy	open	false	2024/4/12	7,000,000,000
databricks/dbrx-instruct	DBRX Instruct	DBRX Instruct	DBRX is a large language model with a fine-grained mixture-of-experts (MoE) architecture that uses 16 experts and chooses 4. It has 132B total parameters, of which 36B parameters are active on any input. ([blog post](https://www.databricks.com/blog/introducing-dbrx-new-state-art-open-llm))	Databricks	open	false	2024/3/27	132,000,000,000
databricks/dolly-v2-12b	Dolly V2 (12B)	null	Dolly V2 (12B) is an instruction-following large language model trained on the Databricks machine learning platform. It is based on pythia-12b.	Databricks	open	true	2023/4/12	11,327,027,200
databricks/dolly-v2-3b	Dolly V2 (3B)	null	Dolly V2 (3B) is an instruction-following large language model trained on the Databricks machine learning platform. It is based on pythia-12b.	Databricks	open	true	2023/4/12	2,517,652,480
databricks/dolly-v2-7b	Dolly V2 (7B)	null	Dolly V2 (7B) is an instruction-following large language model trained on the Databricks machine learning platform. It is based on pythia-12b.	Databricks	open	true	2023/4/12	6,444,163,072
deepmind/chinchilla	Chinchilla (70B)	null	Chinchilla (70B parameters) ([paper](https://arxiv.org/pdf/2203.15556.pdf)).	DeepMind	closed	true	null	null
deepmind/gopher	Gopher (280B)	null	Gopher (540B parameters) ([paper](https://arxiv.org/pdf/2112.11446.pdf)).	DeepMind	closed	true	null	null
deepseek-ai/deepseek-llm-67b-chat	DeepSeek LLM Chat (67B)	DeepSeek LLM Chat (67B)	DeepSeek LLM Chat is a open-source language model trained on 2 trillion tokens in both English and Chinese, and fine-tuned supervised fine-tuning (SFT) and Direct Preference Optimization (DPO). ([paper](https://arxiv.org/abs/2401.02954))	DeepSeek	open	false	2024/1/5	67,000,000,000
eleutherai/pythia-12b-v0	Pythia (12B)	null	Pythia (12B parameters). The Pythia project combines interpretability analysis and scaling laws to understand how knowledge develops and evolves during training in autoregressive transformers.	EleutherAI	open	false	2023/2/13	11,327,027,200
eleutherai/pythia-1b-v0	Pythia (1B)	null	Pythia (1B parameters). The Pythia project combines interpretability analysis and scaling laws to understand how knowledge develops and evolves during training in autoregressive transformers.	EleutherAI	open	true	2023/2/13	805,736,448
eleutherai/pythia-2.8b-v0	Pythia (2.8B)	null	Pythia (2.8B parameters). The Pythia project combines interpretability analysis and scaling laws to understand how knowledge develops and evolves during training in autoregressive transformers.	EleutherAI	open	true	2023/2/13	2,517,652,480
eleutherai/pythia-6.9b	Pythia (6.9B)	null	Pythia (6.9B parameters). The Pythia project combines interpretability analysis and scaling laws to understand how knowledge develops and evolves during training in autoregressive transformers.	EleutherAI	open	false	2023/2/13	6,444,163,072
garage-bAInd/Platypus2-70B-instruct	Platypus2 70B Instruct	Platypus2 70B Instruct	Platypus2 70B Instruct is a large language model trained on 70 billion parameters. ([blog](https://garage-bAInd.com/platypus2-70b-instruct/))	Garage bAInd	open	false	2024/4/18	70,000,000,000
google/code-bison-32k	Codey PaLM-2 (Bison)	null	Codey with a 32K context. PaLM 2 (Pathways Language Model) is a Transformer-based model trained using a mixture of objectives that was evaluated on English and multilingual language, and reasoning tasks. ([report](https://arxiv.org/pdf/2305.10403.pdf))	Google	limited	false	2023/6/29	null
google/code-bison@001	Codey PaLM-2 (Bison)	null	A model fine-tuned to generate code based on a natural language description of the desired code. PaLM 2 (Pathways Language Model) is a Transformer-based model trained using a mixture of objectives that was evaluated on English and multilingual language, and reasoning tasks. ([report](https://arxiv.org/pdf/2305.10403.pdf))	Google	limited	false	2023/6/29	null
google/gemini-1.0-pro-001	Gemini 1.0 Pro (001)	Gemini 1.0 Pro (001)	Gemini 1.0 Pro is a multimodal model able to reason across text, images, video, audio and code. ([paper](https://arxiv.org/abs/2312.11805))	Google	limited	false	2023/12/13	null
google/gemini-1.0-pro-002	Gemini 1.0 Pro (002)	Gemini 1.0 Pro (002)	Gemini 1.0 Pro is a multimodal model able to reason across text, images, video, audio and code. ([paper](https://arxiv.org/abs/2312.11805))	Google	limited	false	2024/4/9	null
google/gemini-1.0-pro-vision-001	Gemini 1.0 Pro Vision	Gemini 1.0 Pro Vision	Gemini 1.0 Pro Vision is a multimodal model able to reason across text, images, video, audio and code. ([paper](https://arxiv.org/abs/2312.11805))	Google	limited	false	2023/12/13	null
google/gemini-1.5-flash-001	Gemini 1.5 Flash (001)	Gemini 1.5 Flash (001)	Gemini 1.5 Flash is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))	Google	limited	false	2024/5/24	null
google/gemini-1.5-flash-001-safety-block-none	Gemini 1.5 Flash (001, BLOCK_NONE safety)	Gemini 1.5 Flash (001, BLOCK_NONE safety)	Gemini 1.5 Flash is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))	Google	limited	false	2024/5/24	null
google/gemini-1.5-flash-002	Gemini 1.5 Flash (002)	Gemini 1.5 Flash (002)	Gemini 1.5 Flash is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))	Google	limited	false	2024/9/24	null
google/gemini-1.5-flash-preview-0514	Gemini 1.5 Flash (0514 preview)	Gemini 1.5 Flash (0514 preview)	Gemini 1.5 Flash is a smaller Gemini model. It has a 1 million token context window and allows interleaving text, images, audio and video as inputs. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([blog](https://blog.google/technology/developers/gemini-gemma-developer-updates-may-2024/))	Google	limited	false	2024/5/14	null
google/gemini-1.5-pro-001	Gemini 1.5 Pro (001)	Gemini 1.5 Pro (001)	Gemini 1.5 Pro is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))	Google	limited	false	2024/5/24	null
google/gemini-1.5-pro-001-safety-block-none	Gemini 1.5 Pro (001, BLOCK_NONE safety)	Gemini 1.5 Pro (001, BLOCK_NONE safety)	Gemini 1.5 Pro is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))	Google	limited	false	2024/5/24	null
google/gemini-1.5-pro-002	Gemini 1.5 Pro (002)	Gemini 1.5 Pro (002)	Gemini 1.5 Pro is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))	Google	limited	false	2024/9/24	null
google/gemini-1.5-pro-preview-0409	Gemini 1.5 Pro (0409 preview)	Gemini 1.5 Pro (0409 preview)	Gemini 1.5 Pro is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))	Google	limited	false	2024/4/10	null
google/gemini-1.5-pro-preview-0514	Gemini 1.5 Pro (0514 preview)	Gemini 1.5 Pro (0514 preview)	Gemini 1.5 Pro is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))	Google	limited	false	2024/5/14	null
google/gemini-2.0-flash-exp	Gemini 2.0 Flash (Experimental)	Gemini 2.0 Flash (Experimental)	Gemini 2.0 Flash (Experimental) is a Gemini model that supports multimodal inputs like images, video and audio, as well as multimodal output like natively generated images mixed with text and steerable text-to-speech (TTS) multilingual audio. ([blog](https://blog.google/technology/google-deepmind/google-gemini-ai-update-december-2024/#gemini-2-0-flash))	Google	limited	false	2024/12/11	null
google/gemma-2-27b	Gemma 2 (27B)	Gemma 2 (27B)	Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/google-gemma-2/))	Google	open	false	2024/6/27	null
google/gemma-2-27b-it	Gemma 2 Instruct (27B)	Gemma 2 Instruct (27B)	Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/google-gemma-2/))	Google	open	false	2024/6/27	null
google/gemma-2-9b	Gemma 2 (9B)	Gemma 2 (9B)	Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/google-gemma-2/))	Google	open	false	2024/6/27	null
google/gemma-2-9b-it	Gemma 2 Instruct (9B)	Gemma 2 Instruct (9B)	Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/google-gemma-2/))	Google	open	false	2024/6/27	null
google/gemma-2b-it	Gemma 2B (IT)	Gemma 2B (IT)	Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/google-gemma-2/))	Google	open	false	2024/6/27	null
google/gemma-7b	Gemma (7B)	Gemma (7B)	Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/gemma-open-models/))	Google	open	false	2024/2/21	null
google/gemma-7b-it	Gemma Instruct (7B)	Gemma Instruct (7B)	TBD	Google	open	false	2024/2/21	null
google/gemma-7b-it	Gemma 7B (IT)	Gemma 7B (IT)	Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/google-gemma-2/))	Google	open	false	2024/6/27	null
google/paligemma-3b-mix-224	PaliGemma (3B) Mix 224	PaliGemma (3B) Mix 224	PaliGemma is a versatile and lightweight vision-language model (VLM) inspired by PaLI-3 and based on open components such as the SigLIP vision model and the Gemma language model. Pre-trained with 224x224 input images and 128 token input/output text sequences. Finetuned on a mixture of downstream academic datasets. ([blog](https://developers.googleblog.com/en/gemma-family-and-toolkit-expansion-io-2024/))	Google	open	false	2024/5/12	null
google/paligemma-3b-mix-448	PaliGemma (3B) Mix 448	PaliGemma (3B) Mix 448	PaliGemma is a versatile and lightweight vision-language model (VLM) inspired by PaLI-3 and based on open components such as the SigLIP vision model and the Gemma language model. Pre-trained with 448x448 input images and 512 token input/output text sequences. Finetuned on a mixture of downstream academic datasets. ([blog](https://developers.googleblog.com/en/gemma-family-and-toolkit-expansion-io-2024/))	Google	open	false	2024/5/12	null
google/palm	PaLM (540B)	null	Pathways Language Model (540B parameters) is trained using 6144 TPU v4 chips ([paper](https://arxiv.org/pdf/2204.02311.pdf)).	Google	closed	true	null	null
google/text-bison-32k	PaLM-2 (Bison)	null	The best value PaLM model with a 32K context. PaLM 2 (Pathways Language Model) is a Transformer-based model trained using a mixture of objectives that was evaluated on English and multilingual language, and reasoning tasks. ([report](https://arxiv.org/pdf/2305.10403.pdf))	Google	limited	false	2023/6/7	null
google/text-bison@001	PaLM-2 (Bison)	null	The best value PaLM model. PaLM 2 (Pathways Language Model) is a Transformer-based model trained using a mixture of objectives that was evaluated on English and multilingual language, and reasoning tasks. ([report](https://arxiv.org/pdf/2305.10403.pdf))	Google	limited	false	2023/6/7	null
google/text-unicorn@001	PaLM-2 (Unicorn)	null	The largest model in PaLM family. PaLM 2 (Pathways Language Model) is a Transformer-based model trained using a mixture of objectives that was evaluated on English and multilingual language, and reasoning tasks. ([report](https://arxiv.org/pdf/2305.10403.pdf))	Google	limited	false	2023/11/30	null
Gryphe/MythoMax-L2-13b	MythoMax L2 13B	MythoMax L2 13B	MythoMax L2 13B is a large language model trained on 13 billion parameters. ([blog](https://gryphe.com/mythomax-l2-13b/))	Gryphe	open	false	2024/4/18	13,000,000,000
huggingface/gpt2	GPT-2 (124M)	null	GPT-2 is a transformers model pretrained on a very large corpus of English data in a self-supervised fashion. This means it was pretrained on the raw texts only, with no humans labelling them in any way (which is why it can use lots of publicly available data) with an automatic process to generate inputs and labels from those texts.	OpenAI	open	false	null	124,000,000
huggingface/gpt2-large	GPT-2 Large (774M)	null	GPT-2 Large is the 774M parameter version of GPT-2, a transformer-based language model created and released by OpenAI. The model is a pretrained model on English language using a causal language modeling (CLM) objective.	OpenAI	open	false	null	774,000,000
huggingface/gpt2-medium	GPT-2 Medium (355M)	null	GPT-2 Medium is the 355M parameter version of GPT-2, a transformer-based language model created and released by OpenAI. The model is a pretrained model on English language using a causal language modeling (CLM) objective.	OpenAI	open	false	null	355,000,000
huggingface/gpt2-xl	GPT-2 XL (1.5B)	null	GPT-2 XL is the 1.5B parameter version of GPT-2, a transformer-based language model created and released by OpenAI. The model is a pretrained model on English language using a causal language modeling (CLM) objective.	OpenAI	open	false	null	1,500,000,000

01-ai/yi-34b

Yi (34B)

null

The Yi models are large language models trained from scratch by developers at 01.AI.

01.AI

open

false

2023/11/2

34,000,000,000

01-ai/yi-34b-chat

Yi Chat (34B)

The Yi models are large language models trained from scratch by developers at 01.AI.

01.AI

open

false

2023/11/23

34,000,000,000

01-ai/yi-6b

Yi (6B)

null

The Yi models are large language models trained from scratch by developers at 01.AI.

01.AI

open

false

2023/11/2

6,000,000,000

01-ai/yi-large-preview

Yi Large (Preview)

The Yi models are large language models trained from scratch by developers at 01.AI. ([tweet](https://x.com/01AI_Yi/status/1789894091620458667))

01.AI

limited

false

2024/5/12

null

ai21/j1-grande

J1-Grande v1 (17B)

null

Jurassic-1 Grande (17B parameters) with a "few tweaks" to the training process ([docs](https://studio.ai21.com/docs/jurassic1-language-models/), [tech report](https://uploads-ssl.webflow.com/60fd4503684b466578c0d307/61138924626a6981ee09caf6_jurassic_tech_paper.pdf)).

AI21 Labs

limited

false

2022/5/3

17,000,000,000

ai21/j1-grande-v2-beta

J1-Grande v2 beta (17B)

null

Jurassic-1 Grande v2 beta (17B parameters)

AI21 Labs

limited

false

2022/10/28

17,000,000,000

ai21/j1-jumbo

J1-Jumbo v1 (178B)

null

Jurassic-1 Jumbo (178B parameters) ([docs](https://studio.ai21.com/docs/jurassic1-language-models/), [tech report](https://uploads-ssl.webflow.com/60fd4503684b466578c0d307/61138924626a6981ee09caf6_jurassic_tech_paper.pdf)).

AI21 Labs

limited

false

2021/8/11

178,000,000,000

ai21/j1-large

J1-Large v1 (7.5B)

null

Jurassic-1 Large (7.5B parameters) ([docs](https://studio.ai21.com/docs/jurassic1-language-models/), [tech report](https://uploads-ssl.webflow.com/60fd4503684b466578c0d307/61138924626a6981ee09caf6_jurassic_tech_paper.pdf)).

AI21 Labs

limited

false

2021/8/11

7,500,000,000

ai21/j2-grande

Jurassic-2 Grande (17B)

null

Jurassic-2 Grande (17B parameters) ([docs](https://www.ai21.com/blog/introducing-j2))

AI21 Labs

limited

false

2023/3/9

17,000,000,000

ai21/j2-jumbo

Jurassic-2 Jumbo (178B)

null

Jurassic-2 Jumbo (178B parameters) ([docs](https://www.ai21.com/blog/introducing-j2))

AI21 Labs

limited

false

2023/3/9

178,000,000,000

ai21/j2-large

Jurassic-2 Large (7.5B)

null

Jurassic-2 Large (7.5B parameters) ([docs](https://www.ai21.com/blog/introducing-j2))

AI21 Labs

limited

false

2023/3/9

7,500,000,000

ai21/jamba-1.5-large

Jamba 1.5 Large

Jamba 1.5 Large is a long-context, hybrid SSM-Transformer instruction following foundation model that is optimized for function calling, structured output, and grounded generation. ([blog](https://www.ai21.com/blog/announcing-jamba-model-family))

AI21 Labs

open

false

2024/8/22

399,000,000,000

ai21/jamba-1.5-mini

Jamba 1.5 Mini

Jamba 1.5 Mini is a long-context, hybrid SSM-Transformer instruction following foundation model that is optimized for function calling, structured output, and grounded generation. ([blog](https://www.ai21.com/blog/announcing-jamba-model-family))

AI21 Labs

open

false

2024/8/22

51,600,000,000

ai21/jamba-instruct

Jamba Instruct

Jamba Instruct is an instruction tuned version of Jamba, which uses a hybrid Transformer-Mamba mixture-of-experts (MoE) architecture that interleaves blocks of Transformer and Mamba layers. ([blog](https://www.ai21.com/blog/announcing-jamba-instruct))

AI21 Labs

limited

false

2024/5/2

52,000,000,000

aisingapore/llama3-8b-cpt-sea-lionv2-base

Llama 3 CPT SEA-Lion v2 (8B)

Llama 3 CPT SEA-Lion v2 (8B) is a multilingual model which was continued pre-trained on 48B additional tokens, including tokens in Southeast Asian languages.

AI Singapore

open

false

2024/7/31

80,300,000,000

aisingapore/llama3-8b-cpt-sea-lionv2.1-instruct

Llama 3 CPT SEA-Lion v2.1 Instruct (8B)

Llama 3 CPT SEA-Lion v2.1 Instruct (8B) is a multilingual model which has been fine-tuned with around 100,000 English instruction-completion pairs alongside a smaller pool of around 50,000 instruction-completion pairs from other Southeast Asian languages, such as Indonesian, Thai and Vietnamese.

AI Singapore

open

false

2024/8/21

80,300,000,000

aisingapore/sea-lion-7b-instruct

SEA-LION Instruct (7B)

SEA-LION is a collection of language models which has been pretrained and instruct-tuned on languages from the Southeast Asia region. It utilizes the MPT architecture and a custom SEABPETokenizer for tokenization.

AI Singapore

open

false

2023/2/24

7,000,000,000

AlephAlpha/luminous-base

Luminous Base (13B)

null

Luminous Base (13B parameters) ([docs](https://docs.aleph-alpha.com/docs/introduction/luminous/))

Aleph Alpha

limited

false

2022/1/1

13,000,000,000

AlephAlpha/luminous-extended

Luminous Extended (30B)

null

Luminous Extended (30B parameters) ([docs](https://docs.aleph-alpha.com/docs/introduction/luminous/))

Aleph Alpha

limited

false

2022/1/1

30,000,000,000

AlephAlpha/luminous-supreme

Luminous Supreme (70B)

null

Luminous Supreme (70B parameters) ([docs](https://docs.aleph-alpha.com/docs/introduction/luminous/))

Aleph Alpha

limited

false

2022/1/1

70,000,000,000

allenai/olmo-1.7-7b

OLMo 1.7 (7B)

OLMo is a series of Open Language Models trained on the Dolma dataset. The instruct versions was trained on the Tulu SFT mixture and a cleaned version of the UltraFeedback dataset.

Allen Institute for AI

open

false

2024/4/17

7,000,000,000

allenai/olmo-7b

OLMo (7B)

OLMo is a series of Open Language Models trained on the Dolma dataset.

Allen Institute for AI

open

false

2024/2/1

7,000,000,000

anthropic/claude-2.0

Anthropic Claude 2.0

null

Claude 2.0 is a general purpose large language model developed by Anthropic. It uses a transformer architecture and is trained via unsupervised learning, RLHF, and Constitutional AI (including both a supervised and Reinforcement Learning (RL) phase). ([model card](https://efficient-manatee.files.svdcdn.com/production/images/Model-Card-Claude-2.pdf))

Anthropic

limited

false

2023/7/11

null

anthropic/claude-2.1

Anthropic Claude 2.1

null

Claude 2.1 is a general purpose large language model developed by Anthropic. It uses a transformer architecture and is trained via unsupervised learning, RLHF, and Constitutional AI (including both a supervised and Reinforcement Learning (RL) phase). ([model card](https://efficient-manatee.files.svdcdn.com/production/images/Model-Card-Claude-2.pdf))

Anthropic

limited

false

2023/11/21

null

anthropic/claude-3-5-haiku-20241022

Claude 3.5 Haiku (20241022)

Claude 3.5 Haiku is a Claude 3 family model which matches the performance of Claude 3 Opus at a similar speed to the previous generation of Haiku ([blog](https://www.anthropic.com/news/3-5-models-and-computer-use)).

Anthropic

limited

false

2024/11/4

null

anthropic/claude-3-5-sonnet-20240620

Claude 3.5 Sonnet (20240620)

Claude 3.5 Sonnet is a Claude 3 family model which outperforms Claude 3 Opus while operating faster and at a lower cost. ([blog](https://www.anthropic.com/news/claude-3-5-sonnet))

Anthropic

limited

false

2024/6/20

null

anthropic/claude-3-5-sonnet-20241022

Claude 3.5 Sonnet (20241022)

Claude 3.5 Sonnet is a Claude 3 family model which outperforms Claude 3 Opus while operating faster and at a lower cost ([blog](https://www.anthropic.com/news/claude-3-5-sonnet)). This is an upgraded snapshot released on 2024-10-22 ([blog](https://www.anthropic.com/news/3-5-models-and-computer-use)).

Anthropic

limited

false

2024/10/22

null

anthropic/claude-3-haiku-20240307

Claude 3 Haiku (20240307)

Claude 3 is a a family of models that possess vision and multilingual capabilities. They were trained with various methods such as unsupervised learning and Constitutional AI ([blog](https://www.anthropic.com/news/claude-3-family)).

Anthropic

limited

false

2024/3/13

null

anthropic/claude-3-opus-20240229

Claude 3 Opus (20240229)

Claude 3 is a a family of models that possess vision and multilingual capabilities. They were trained with various methods such as unsupervised learning and Constitutional AI ([blog](https://www.anthropic.com/news/claude-3-family)).

Anthropic

limited

false

2024/3/4

null

anthropic/claude-3-sonnet-20240229

Claude 3 Sonnet (20240229)

Claude 3 is a a family of models that possess vision and multilingual capabilities. They were trained with various methods such as unsupervised learning and Constitutional AI ([blog](https://www.anthropic.com/news/claude-3-family)).

Anthropic

limited

false

2024/3/4

null

anthropic/claude-instant-1.2

Anthropic Claude Instant 1.2

null

A lightweight version of Claude, a model trained using reinforcement learning from human feedback ([docs](https://www.anthropic.com/index/introducing-claude)).

Anthropic

limited

false

2023/8/9

null

anthropic/claude-instant-v1

Anthropic Claude Instant V1

null

A lightweight version of Claude, a model trained using reinforcement learning from human feedback ([docs](https://www.anthropic.com/index/introducing-claude)).

Anthropic

limited

false

2023/3/17

null

anthropic/claude-v1.3

Anthropic Claude v1.3

null

A model trained using reinforcement learning from human feedback ([docs](https://www.anthropic.com/index/introducing-claude)).

Anthropic

limited

false

2023/3/17

null

anthropic/stanford-online-all-v4-s3

Anthropic-LM v4-s3 (52B)

null

A 52B parameter language model, trained using reinforcement learning from human feedback [paper](https://arxiv.org/pdf/2204.05862.pdf).

Anthropic

closed

false

2021/12/1

52,000,000,000

Austism/chronos-hermes-13b

Chronos Hermes 13B

Chronos Hermes 13B is a large language model trained on 13 billion parameters. ([blog](https://chronos.ai/chronos-hermes-13b/))

Chronos

open

false

2024/4/18

13,000,000,000

codellama/CodeLlama-13b-Instruct-hf

CodeLlama 13B Instruct

CodeLlama 13B Instruct is a large language model trained on 13 billion parameters. ([blog](https://codellama.com/codellama-13b-instruct/))

CodeLlama

open

false

2024/4/18

13,000,000,000

codellama/CodeLlama-34b-Instruct-hf

CodeLlama 34B Instruct

CodeLlama 34B Instruct is a large language model trained on 34 billion parameters. ([blog](https://codellama.com/codellama-34b-instruct/))

CodeLlama

open

false

2024/4/18

34,000,000,000

codellama/CodeLlama-70b-Instruct-hf

CodeLlama 70B Instruct

CodeLlama 70B Instruct is a large language model trained on 70 billion parameters. ([blog](https://codellama.com/codellama-70b-instruct/))

CodeLlama

open

false

2024/4/18

70,000,000,000

codellama/CodeLlama-7b-Instruct-hf

CodeLlama 7B Instruct

CodeLlama 7B Instruct is a large language model trained on 7 billion parameters. ([blog](https://codellama.com/codellama-7b-instruct/))

CodeLlama

open

false

2024/4/18

7,000,000,000

cognitivecomputations/dolphin-2.5-mixtral-8x7b

Dolphin 2.5 Mixtral 8x7B

Dolphin 2.5 Mixtral 8x7B is a multimodal model trained on 8x7B parameters with a 32K token sequence length. ([blog](https://cognitivecomputations.com/dolphin-2.5-mixtral-8x7b/))

Cognitive Computations

open

false

2024/4/18

46,700,000,000

cohere/command

Cohere Command

null

Command is Cohere’s flagship text generation model. It is trained to follow user commands and to be instantly useful in practical business applications. [docs](https://docs.cohere.com/reference/generate) and [changelog](https://docs.cohere.com/changelog)

Cohere

limited

false

2023/9/29

null

cohere/command-light

Cohere Command Light

null

Command is Cohere’s flagship text generation model. It is trained to follow user commands and to be instantly useful in practical business applications. [docs](https://docs.cohere.com/reference/generate) and [changelog](https://docs.cohere.com/changelog)

Cohere

limited

false

2023/9/29

null

cohere/command-medium-beta

Cohere Command beta (6.1B)

null

Cohere Command beta (6.1B parameters) is fine-tuned from the medium model to respond well with instruction-like prompts ([details](https://docs.cohere.ai/docs/command-beta)).

Cohere

limited

false

2022/11/8

6,100,000,000

cohere/command-r

Command R

Command R is a multilingual 35B parameter model with a context length of 128K that has been trained with conversational tool use capabilities.

Cohere

open

false

2024/3/11

35,000,000,000

cohere/command-r-plus

Command R Plus

Command R+ is a multilingual 104B parameter model with a context length of 128K that has been trained with conversational tool use capabilities.

Cohere

open

false

2024/4/4

104,000,000,000

cohere/command-xlarge-beta

Cohere Command beta (52.4B)

null

Cohere Command beta (52.4B parameters) is fine-tuned from the XL model to respond well with instruction-like prompts ([details](https://docs.cohere.ai/docs/command-beta)).

Cohere

limited

false

2022/11/8

52,400,000,000

cohere/large-20220720

Cohere large v20220720 (13.1B)

null

Cohere large v20220720 (13.1B parameters), which is deprecated by Cohere as of December 2, 2022.

Cohere

limited

false

2022/7/20

13,100,000,000

cohere/medium-20220720

Cohere medium v20220720 (6.1B)

null

Cohere medium v20220720 (6.1B parameters)

Cohere

limited

false

2022/7/20

6,100,000,000

cohere/medium-20221108

Cohere medium v20221108 (6.1B)

null

Cohere medium v20221108 (6.1B parameters)

Cohere

limited

false

2022/11/8

6,100,000,000

cohere/small-20220720

Cohere small v20220720 (410M)

null

Cohere small v20220720 (410M parameters), which is deprecated by Cohere as of December 2, 2022.

Cohere

limited

false

2022/7/20

410,000,000

cohere/xlarge-20220609

Cohere xlarge v20220609 (52.4B)

null

Cohere xlarge v20220609 (52.4B parameters)

Cohere

limited

false

2022/6/9

52,400,000,000

cohere/xlarge-20221108

Cohere xlarge v20221108 (52.4B)

null

Cohere xlarge v20221108 (52.4B parameters)

Cohere

limited

false

2022/11/8

52,400,000,000

damo/seallm-7b-v2

SeaLLM v2 (7B)

SeaLLM v2 is a multilingual LLM for Southeast Asian (SEA) languages trained from Mistral (7B). ([website](https://damo-nlp-sg.github.io/SeaLLMs/))

Alibaba DAMO Academy

open

false

2024/2/2

7,000,000,000

damo/seallm-7b-v2.5

SeaLLM v2.5 (7B)

SeaLLM is a multilingual LLM for Southeast Asian (SEA) languages trained from Gemma (7B). ([website](https://damo-nlp-sg.github.io/SeaLLMs/))

Alibaba DAMO Academy

open

false

2024/4/12

7,000,000,000

databricks/dbrx-instruct

DBRX Instruct

DBRX is a large language model with a fine-grained mixture-of-experts (MoE) architecture that uses 16 experts and chooses 4. It has 132B total parameters, of which 36B parameters are active on any input. ([blog post](https://www.databricks.com/blog/introducing-dbrx-new-state-art-open-llm))

Databricks

open

false

2024/3/27

132,000,000,000

databricks/dolly-v2-12b

Dolly V2 (12B)

null

Dolly V2 (12B) is an instruction-following large language model trained on the Databricks machine learning platform. It is based on pythia-12b.

Databricks

open

true

2023/4/12

11,327,027,200

databricks/dolly-v2-3b

Dolly V2 (3B)

null

Dolly V2 (3B) is an instruction-following large language model trained on the Databricks machine learning platform. It is based on pythia-12b.

Databricks

open

true

2023/4/12

2,517,652,480

databricks/dolly-v2-7b

Dolly V2 (7B)

null

Dolly V2 (7B) is an instruction-following large language model trained on the Databricks machine learning platform. It is based on pythia-12b.

Databricks

open

true

2023/4/12

6,444,163,072

deepmind/chinchilla

Chinchilla (70B)

null

Chinchilla (70B parameters) ([paper](https://arxiv.org/pdf/2203.15556.pdf)).

DeepMind

closed

true

null

deepmind/gopher

Gopher (280B)

null

Gopher (540B parameters) ([paper](https://arxiv.org/pdf/2112.11446.pdf)).

DeepMind

closed

true

null

deepseek-ai/deepseek-llm-67b-chat

DeepSeek LLM Chat (67B)

DeepSeek LLM Chat is a open-source language model trained on 2 trillion tokens in both English and Chinese, and fine-tuned supervised fine-tuning (SFT) and Direct Preference Optimization (DPO). ([paper](https://arxiv.org/abs/2401.02954))

DeepSeek

open

false

2024/1/5

67,000,000,000

eleutherai/pythia-12b-v0

Pythia (12B)

null

Pythia (12B parameters). The Pythia project combines interpretability analysis and scaling laws to understand how knowledge develops and evolves during training in autoregressive transformers.

EleutherAI

open

false

2023/2/13

11,327,027,200

eleutherai/pythia-1b-v0

Pythia (1B)

null

Pythia (1B parameters). The Pythia project combines interpretability analysis and scaling laws to understand how knowledge develops and evolves during training in autoregressive transformers.

EleutherAI

open

true

2023/2/13

805,736,448

eleutherai/pythia-2.8b-v0

Pythia (2.8B)

null

Pythia (2.8B parameters). The Pythia project combines interpretability analysis and scaling laws to understand how knowledge develops and evolves during training in autoregressive transformers.

EleutherAI

open

true

2023/2/13

2,517,652,480

eleutherai/pythia-6.9b

Pythia (6.9B)

null

Pythia (6.9B parameters). The Pythia project combines interpretability analysis and scaling laws to understand how knowledge develops and evolves during training in autoregressive transformers.

EleutherAI

open

false

2023/2/13

6,444,163,072

garage-bAInd/Platypus2-70B-instruct

Platypus2 70B Instruct

Platypus2 70B Instruct is a large language model trained on 70 billion parameters. ([blog](https://garage-bAInd.com/platypus2-70b-instruct/))

Garage bAInd

open

false

2024/4/18

70,000,000,000

google/code-bison-32k

Codey PaLM-2 (Bison)

null

Codey with a 32K context. PaLM 2 (Pathways Language Model) is a Transformer-based model trained using a mixture of objectives that was evaluated on English and multilingual language, and reasoning tasks. ([report](https://arxiv.org/pdf/2305.10403.pdf))

Google

limited

false

2023/6/29

null

google/code-bison@001

Codey PaLM-2 (Bison)

null

A model fine-tuned to generate code based on a natural language description of the desired code. PaLM 2 (Pathways Language Model) is a Transformer-based model trained using a mixture of objectives that was evaluated on English and multilingual language, and reasoning tasks. ([report](https://arxiv.org/pdf/2305.10403.pdf))

Google

limited

false

2023/6/29

null

google/gemini-1.0-pro-001

Gemini 1.0 Pro (001)

Gemini 1.0 Pro is a multimodal model able to reason across text, images, video, audio and code. ([paper](https://arxiv.org/abs/2312.11805))

Google

limited

false

2023/12/13

null

google/gemini-1.0-pro-002

Gemini 1.0 Pro (002)

Gemini 1.0 Pro is a multimodal model able to reason across text, images, video, audio and code. ([paper](https://arxiv.org/abs/2312.11805))

Google

limited

false

2024/4/9

null

google/gemini-1.0-pro-vision-001

Gemini 1.0 Pro Vision

Gemini 1.0 Pro Vision is a multimodal model able to reason across text, images, video, audio and code. ([paper](https://arxiv.org/abs/2312.11805))

Google

limited

false

2023/12/13

null

google/gemini-1.5-flash-001

Gemini 1.5 Flash (001)

Gemini 1.5 Flash is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))

Google

limited

false

2024/5/24

null

google/gemini-1.5-flash-001-safety-block-none

Gemini 1.5 Flash (001, BLOCK_NONE safety)

Gemini 1.5 Flash is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))

Google

limited

false

2024/5/24

null

google/gemini-1.5-flash-002

Gemini 1.5 Flash (002)

Gemini 1.5 Flash is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))

Google

limited

false

2024/9/24

null

google/gemini-1.5-flash-preview-0514

Gemini 1.5 Flash (0514 preview)

Gemini 1.5 Flash is a smaller Gemini model. It has a 1 million token context window and allows interleaving text, images, audio and video as inputs. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([blog](https://blog.google/technology/developers/gemini-gemma-developer-updates-may-2024/))

Google

limited

false

2024/5/14

null

google/gemini-1.5-pro-001

Gemini 1.5 Pro (001)

Gemini 1.5 Pro is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))

Google

limited

false

2024/5/24

null

google/gemini-1.5-pro-001-safety-block-none

Gemini 1.5 Pro (001, BLOCK_NONE safety)

Gemini 1.5 Pro is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))

Google

limited

false

2024/5/24

null

google/gemini-1.5-pro-002

Gemini 1.5 Pro (002)

Gemini 1.5 Pro is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))

Google

limited

false

2024/9/24

null

google/gemini-1.5-pro-preview-0409

Gemini 1.5 Pro (0409 preview)

Gemini 1.5 Pro is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))

Google

limited

false

2024/4/10

null

google/gemini-1.5-pro-preview-0514

Gemini 1.5 Pro (0514 preview)

Gemini 1.5 Pro is a multimodal mixture-of-experts model capable of recalling and reasoning over fine-grained information from long contexts. This model is accessed through Vertex AI and has all safety thresholds set to `BLOCK_NONE`. ([paper](https://arxiv.org/abs/2403.05530))

Google

limited

false

2024/5/14

null

google/gemini-2.0-flash-exp

Gemini 2.0 Flash (Experimental)

Gemini 2.0 Flash (Experimental) is a Gemini model that supports multimodal inputs like images, video and audio, as well as multimodal output like natively generated images mixed with text and steerable text-to-speech (TTS) multilingual audio. ([blog](https://blog.google/technology/google-deepmind/google-gemini-ai-update-december-2024/#gemini-2-0-flash))

Google

limited

false

2024/12/11

null

google/gemma-2-27b

Gemma 2 (27B)

Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/google-gemma-2/))

Google

open

false

2024/6/27

null

google/gemma-2-27b-it

Gemma 2 Instruct (27B)

Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/google-gemma-2/))

Google

open

false

2024/6/27

null

google/gemma-2-9b

Gemma 2 (9B)

Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/google-gemma-2/))

Google

open

false

2024/6/27

null

google/gemma-2-9b-it

Gemma 2 Instruct (9B)

Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/google-gemma-2/))

Google

open

false

2024/6/27

null

google/gemma-2b-it

Gemma 2B (IT)

Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/google-gemma-2/))

Google

open

false

2024/6/27

null

google/gemma-7b

Gemma (7B)

Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/gemma-open-models/))

Google

open

false

2024/2/21

null

google/gemma-7b-it

Gemma Instruct (7B)

TBD

Google

open

false

2024/2/21

null

google/gemma-7b-it

Gemma 7B (IT)

Gemma is a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. ([model card](https://www.kaggle.com/models/google/gemma), [blog post](https://blog.google/technology/developers/google-gemma-2/))

Google

open

false

2024/6/27

null

google/paligemma-3b-mix-224

PaliGemma (3B) Mix 224

PaliGemma is a versatile and lightweight vision-language model (VLM) inspired by PaLI-3 and based on open components such as the SigLIP vision model and the Gemma language model. Pre-trained with 224x224 input images and 128 token input/output text sequences. Finetuned on a mixture of downstream academic datasets. ([blog](https://developers.googleblog.com/en/gemma-family-and-toolkit-expansion-io-2024/))

Google

open

false

2024/5/12

null

google/paligemma-3b-mix-448

PaliGemma (3B) Mix 448

PaliGemma is a versatile and lightweight vision-language model (VLM) inspired by PaLI-3 and based on open components such as the SigLIP vision model and the Gemma language model. Pre-trained with 448x448 input images and 512 token input/output text sequences. Finetuned on a mixture of downstream academic datasets. ([blog](https://developers.googleblog.com/en/gemma-family-and-toolkit-expansion-io-2024/))

Google

open

false

2024/5/12

null

google/palm

PaLM (540B)

null

Pathways Language Model (540B parameters) is trained using 6144 TPU v4 chips ([paper](https://arxiv.org/pdf/2204.02311.pdf)).

Google

closed

true

null

google/text-bison-32k

PaLM-2 (Bison)

null

The best value PaLM model with a 32K context. PaLM 2 (Pathways Language Model) is a Transformer-based model trained using a mixture of objectives that was evaluated on English and multilingual language, and reasoning tasks. ([report](https://arxiv.org/pdf/2305.10403.pdf))

Google

limited

false

2023/6/7

null

google/text-bison@001

PaLM-2 (Bison)

null

The best value PaLM model. PaLM 2 (Pathways Language Model) is a Transformer-based model trained using a mixture of objectives that was evaluated on English and multilingual language, and reasoning tasks. ([report](https://arxiv.org/pdf/2305.10403.pdf))

Google

limited

false

2023/6/7

null

google/text-unicorn@001

PaLM-2 (Unicorn)

null

The largest model in PaLM family. PaLM 2 (Pathways Language Model) is a Transformer-based model trained using a mixture of objectives that was evaluated on English and multilingual language, and reasoning tasks. ([report](https://arxiv.org/pdf/2305.10403.pdf))

Google

limited

false

2023/11/30

null

Gryphe/MythoMax-L2-13b

MythoMax L2 13B

MythoMax L2 13B is a large language model trained on 13 billion parameters. ([blog](https://gryphe.com/mythomax-l2-13b/))

Gryphe

open

false

2024/4/18

13,000,000,000

huggingface/gpt2

GPT-2 (124M)

null

GPT-2 is a transformers model pretrained on a very large corpus of English data in a self-supervised fashion. This means it was pretrained on the raw texts only, with no humans labelling them in any way (which is why it can use lots of publicly available data) with an automatic process to generate inputs and labels from those texts.

OpenAI

open

false

null

124,000,000

huggingface/gpt2-large

GPT-2 Large (774M)

null

GPT-2 Large is the 774M parameter version of GPT-2, a transformer-based language model created and released by OpenAI. The model is a pretrained model on English language using a causal language modeling (CLM) objective.

OpenAI

open

false

null

774,000,000

huggingface/gpt2-medium

GPT-2 Medium (355M)

null

GPT-2 Medium is the 355M parameter version of GPT-2, a transformer-based language model created and released by OpenAI. The model is a pretrained model on English language using a causal language modeling (CLM) objective.

OpenAI

open

false

null

355,000,000

huggingface/gpt2-xl

GPT-2 XL (1.5B)

null

GPT-2 XL is the 1.5B parameter version of GPT-2, a transformer-based language model created and released by OpenAI. The model is a pretrained model on English language using a causal language modeling (CLM) objective.

OpenAI

open

false

null

1,500,000,000