- Llama 3 Lumimaid 70B
The NeverSleep team is back, with a Llama 3 70B finetune trained on their curated roleplay data. Striking a balance between eRP and RP, Lumimaid was designed to be serious, yet uncensored when necessary. To enhance it's overall intelligence and chat capability, roughly 40% of the training data was not roleplay. This provides a breadth of knowledge to access, while still keeping roleplay as the primary strength. Usage of this model is subject to Meta's Acceptable Use Policy.
by neversleep8K context$8/M input tkns$8/M output tkns7.33M tokens this week - Google: Gemini Flash 1.5 (preview)
Gemini 1.5 Flash is a foundation model that performs well at a variety of multimodal tasks such as visual understanding, classification, summarization, and creating content from image, audio and video. It's adept at processing visual and text inputs such as photographs, documents, infographics, and screenshots. Gemini 1.5 Flash is designed for high-volume, high-frequency tasks where cost and latency matter. On most common tasks, Flash achieves comparable quality to other Gemini Pro models at a significantly reduced cost. Flash is well-suited for applications like chat assistants and on-demand content generation where speed and scale matter. #multimodal
by google2.8M context$0.25/M input tkns$0.75/M output tkns$0.265/K input imgs196M tokens this week - Perplexity: Llama3 Sonar 8B
Llama3 Sonar is Perplexity's latest model family. It surpasses their earlier Sonar models in cost-efficiency, speed, and performance. This is a normal offline LLM, but the online version of this model has Internet access.
by perplexity33K context$0.2/M input tkns$0.2/M output tkns16.2M tokens this week - Perplexity: Llama3 Sonar 8B Online
Llama3 Sonar is Perplexity's latest model family. It surpasses their earlier Sonar models in cost-efficiency, speed, and performance. This is the online version of the offline chat model. It is focused on delivering helpful, up-to-date, and factual responses. #online
by perplexity28K context$0.2/M input tkns$0.2/M output tkns$5/K reqs2.39M tokens this week - Perplexity: Llama3 Sonar 70B
Llama3 Sonar is Perplexity's latest model family. It surpasses their earlier Sonar models in cost-efficiency, speed, and performance. This is a normal offline LLM, but the online version of this model has Internet access.
by perplexity33K context$1/M input tkns$1/M output tkns15M tokens this week - Perplexity: Llama3 Sonar 70B Online
Llama3 Sonar is Perplexity's latest model family. It surpasses their earlier Sonar models in cost-efficiency, speed, and performance. This is the online version of the offline chat model. It is focused on delivering helpful, up-to-date, and factual responses. #online
by perplexity28K context$1/M input tkns$1/M output tkns$5/K reqs11.9M tokens this week - DeepSeek-V2 Chat
DeepSeek-V2 Chat is a conversational finetune of DeepSeek-V2, a Mixture-of-Experts (MoE) language model. It comprises 236B total parameters, of which 21B are activated for each token. Compared with DeepSeek 67B, DeepSeek-V2 achieves stronger performance, and meanwhile saves 42.5% of training costs, reduces the KV cache by 93.3%, and boosts the maximum generation throughput to 5.76 times. DeepSeek-V2 achieves remarkable performance on both standard benchmarks and open-ended generation evaluations.
by deepseek33K context$0.14/M input tkns$0.28/M output tkns51.3M tokens this week - Deepseek Coder
Deepseek Coder is composed of a series of code language models, each trained from scratch on 2T tokens, with a composition of 87% code and 13% natural language in both English and Chinese. The model is pre-trained on project-level code corpus by employing a window size of 16K and a extra fill-in-the-blank task, to support project-level code completion and infilling. For coding capabilities, Deepseek Coder achieves state-of-the-art performance among open-source code models on multiple programming languages and various benchmarks
by deepseek16K context$0.14/M input tkns$0.28/M output tkns12.6M tokens this week - Meta: Llama 3 8B
Meta's latest class of model (Llama 3) launched with a variety of sizes & flavors. This is the base 8B pre-trained version. It has demonstrated strong performance compared to leading closed-source models in human evaluations. To read more about the model release, click here. Usage of this model is subject to Meta's Acceptable Use Policy.
by meta-llama8K context$0.18/M input tkns$0.18/M output tkns3.65M tokens this week - Meta: Llama 3 70B
Meta's latest class of model (Llama 3) launched with a variety of sizes & flavors. This is the base 70B pre-trained version. It has demonstrated strong performance compared to leading closed-source models in human evaluations. To read more about the model release, click here. Usage of this model is subject to Meta's Acceptable Use Policy.
by meta-llama8K context$0.81/M input tkns$0.81/M output tkns7.27M tokens this week - Meta: LlamaGuard 2 8B
This safeguard model has 8B parameters and is based on the Llama 3 family. Just like is predecessor, LlamaGuard 1, it can do both prompt and response classification. LlamaGuard 2 acts as a normal LLM would, generating text that indicates whether the given input/output is safe/unsafe. If deemed unsafe, it will also share the content categories violated. For best results, please use raw prompt input or the
/completions
endpoint, instead of the chat API. It has demonstrated strong performance compared to leading closed-source models in human evaluations. To read more about the model release, click here. Usage of this model is subject to Meta's Acceptable Use Policy.by meta-llama8K context$0.18/M input tkns$0.18/M output tkns255K tokens this week - OpenAI: GPT-4o
GPT-4o ("o" for "omni") is OpenAI's latest AI model, supporting both text and image inputs with text outputs. It maintains the intelligence level of GPT-4 Turbo while being twice as fast and 50% more cost-effective. GPT-4o also offers improved performance in processing non-English languages and enhanced visual capabilities. For benchmarking against other models, it was briefly called "im-also-a-good-gpt2-chatbot" #multimodal
by openai128K context$5/M input tkns$15/M output tkns$2.312/K input imgs1.49B tokens this week - OpenAI: GPT-4o (2024-05-13)
GPT-4o ("o" for "omni") is OpenAI's latest AI model, supporting both text and image inputs with text outputs. It maintains the intelligence level of GPT-4 Turbo while being twice as fast and 50% more cost-effective. GPT-4o also offers improved performance in processing non-English languages and enhanced visual capabilities. For benchmarking against other models, it was briefly called "im-also-a-good-gpt2-chatbot" #multimodal
by openai128K context$5/M input tkns$15/M output tkns$2.312/K input imgs139M tokens this week - LLaVA v1.6 34B
LLaVA Yi 34B is an open-source model trained by fine-tuning LLM on multimodal instruction-following data. It is an auto-regressive language model, based on the transformer architecture. Base LLM: NousResearch/Nous-Hermes-2-Yi-34B It was trained in December 2023.
by liuhaotian4K context$0.9/M input tkns$0.9/M output tkns$0.5184/K input imgs2.53M tokens this week - OLMo 7B Instruct
OLMo 7B Instruct by the Allen Institute for AI is a model finetuned for question answering. It demonstrates notable performance across multiple benchmarks including TruthfulQA and ToxiGen. Open Source: The model, its code, checkpoints, logs are released under the Apache 2.0 license. - Core repo (training, inference, fine-tuning etc.) - Evaluation code - Further fine-tuning code - Paper - Technical blog post - W&B Logs
by allenai2K context$0.18/M input tkns$0.18/M output tkns523K tokens this week - Qwen 1.5 110B Chat
Qwen1.5 110B is the beta version of Qwen2, a transformer-based decoder-only language model pretrained on a large amount of data. In comparison with the previous released Qwen, the improvements include: - Significant performance improvement in human preference for chat models - Multilingual support of both base and chat models - Stable support of 32K context length for models of all sizes For more details, see this blog post and GitHub repo. Usage of this model is subject to Tongyi Qianwen LICENSE AGREEMENT.
by qwen33K context$1.62/M input tkns$1.62/M output tkns19.5M tokens this week - Qwen 1.5 32B Chat
Qwen1.5 32B is the beta version of Qwen2, a transformer-based decoder-only language model pretrained on a large amount of data. In comparison with the previous released Qwen, the improvements include: - Significant performance improvement in human preference for chat models - Multilingual support of both base and chat models - Stable support of 32K context length for models of all sizes For more details, see this blog post and GitHub repo. Usage of this model is subject to Tongyi Qianwen LICENSE AGREEMENT.
by qwen33K context$0.72/M input tkns$0.72/M output tkns5.57M tokens this week - Qwen 1.5 14B Chat
Qwen1.5 14B is the beta version of Qwen2, a transformer-based decoder-only language model pretrained on a large amount of data. In comparison with the previous released Qwen, the improvements include: - Significant performance improvement in human preference for chat models - Multilingual support of both base and chat models - Stable support of 32K context length for models of all sizes For more details, see this blog post and GitHub repo. Usage of this model is subject to Tongyi Qianwen LICENSE AGREEMENT.
by qwen33K context$0.27/M input tkns$0.27/M output tkns31M tokens this week - Qwen 1.5 7B Chat
Qwen1.5 7B is the beta version of Qwen2, a transformer-based decoder-only language model pretrained on a large amount of data. In comparison with the previous released Qwen, the improvements include: - Significant performance improvement in human preference for chat models - Multilingual support of both base and chat models - Stable support of 32K context length for models of all sizes For more details, see this blog post and GitHub repo. Usage of this model is subject to Tongyi Qianwen LICENSE AGREEMENT.
by qwen33K context$0.18/M input tkns$0.18/M output tkns5.16M tokens this week - Qwen 1.5 4B Chat
Qwen1.5 4B is the beta version of Qwen2, a transformer-based decoder-only language model pretrained on a large amount of data. In comparison with the previous released Qwen, the improvements include: - Significant performance improvement in human preference for chat models - Multilingual support of both base and chat models - Stable support of 32K context length for models of all sizes For more details, see this blog post and GitHub repo. Usage of this model is subject to Tongyi Qianwen LICENSE AGREEMENT.
by qwen33K context$0.09/M input tkns$0.09/M output tkns883K tokens this week - Qwen 1.5 72B Chat
Qwen1.5 72B is the beta version of Qwen2, a transformer-based decoder-only language model pretrained on a large amount of data. In comparison with the previous released Qwen, the improvements include: - Significant performance improvement in human preference for chat models - Multilingual support of both base and chat models - Stable support of 32K context length for models of all sizes For more details, see this blog post and GitHub repo. Usage of this model is subject to Tongyi Qianwen LICENSE AGREEMENT.
by qwen33K context$0.9/M input tkns$0.9/M output tkns30.4M tokens this week - Meta: Llama 3 8B Instruct (free)
Meta's latest class of model (Llama 3) launched with a variety of sizes & flavors. This 8B instruct-tuned version was optimized for high quality dialogue usecases. It has demonstrated strong performance compared to leading closed-source models in human evaluations. To read more about the model release, click here. Usage of this model is subject to Meta's Acceptable Use Policy. Note: this is a free, rate-limited version of this model. Outputs may be cached. Read about rate limits here.
by meta-llama8K context$0/M input tkns$0/M output tkns45.8M tokens this week - Llama 3 Lumimaid 8B
The NeverSleep team is back, with a Llama 3 8B finetune trained on their curated roleplay data. Striking a balance between eRP and RP, Lumimaid was designed to be serious, yet uncensored when necessary. To enhance it's overall intelligence and chat capability, roughly 40% of the training data was not roleplay. This provides a breadth of knowledge to access, while still keeping roleplay as the primary strength. Usage of this model is subject to Meta's Acceptable Use Policy.
by neversleep25K context$0.225/M input tkns$2.25/M output tkns127M tokens this week - Llama 3 Lumimaid 8B (extended)
The NeverSleep team is back, with a Llama 3 8B finetune trained on their curated roleplay data. Striking a balance between eRP and RP, Lumimaid was designed to be serious, yet uncensored when necessary. To enhance it's overall intelligence and chat capability, roughly 40% of the training data was not roleplay. This provides a breadth of knowledge to access, while still keeping roleplay as the primary strength. Usage of this model is subject to Meta's Acceptable Use Policy. Note: this is an extended-context version of this model. It may have higher prices and different outputs.
by neversleep25K context$0.225/M input tkns$2.25/M output tkns102M tokens this week - Snowflake: Arctic Instruct
Arctic is a dense-MoE Hybrid transformer architecture pre-trained from scratch by the Snowflake AI Research Team. Arctic combines a 10B dense transformer model with a residual 128x3.66B MoE MLP resulting in 480B total and 17B active parameters chosen using a top-2 gating. To read more about this model's release, click here.
by snowflake4K context$2.16/M input tkns$2.16/M output tkns880K tokens this week - FireLLaVA 13B
A blazing fast vision-language model, FireLLaVA quickly understands both text and images. It achieves impressive chat skills in tests, and was designed to mimic multimodal GPT-4. The first commercially permissive open source LLaVA model, trained entirely on open source LLM generated instruction following data.
by fireworks4K context$0.2/M input tkns$0.2/M output tkns$0.1152/K input imgs8.27M tokens this week - Lynn: Llama 3 Soliloquy 8B v2
Soliloquy-L3 v2 is a fast, highly capable roleplaying model designed for immersive, dynamic experiences. Trained on over 250 million tokens of roleplaying data, Soliloquy-L3 has a vast knowledge base, rich literary expression, and support for up to 24k context length. It outperforms existing ~13B models, delivering enhanced roleplaying capabilities. Usage of this model is subject to Meta's Acceptable Use Policy.
by lynn25K context$0.05/M input tkns$0.05/M output tkns53.7M tokens this week - Fimbulvetr 11B v2
Creative writing model, routed with permission. It's fast, it keeps the conversation going, and it stays in character. If you submit a raw prompt, you can use Alpaca or Vicuna formats.
by sao10k8K context$0.5499/M input tkns$2.826/M output tkns44.4M tokens this week - Meta: Llama 3 8B Instruct (extended)
Meta's latest class of model (Llama 3) launched with a variety of sizes & flavors. This 8B instruct-tuned version was optimized for high quality dialogue usecases. It has demonstrated strong performance compared to leading closed-source models in human evaluations. To read more about the model release, click here. Usage of this model is subject to Meta's Acceptable Use Policy. Note: this is an extended-context version of this model. It may have higher prices and different outputs.
by meta-llama16K context$0.225/M input tkns$2.25/M output tkns14.2M tokens this week - Meta: Llama 3 8B Instruct (nitro)
Meta's latest class of model (Llama 3) launched with a variety of sizes & flavors. This 8B instruct-tuned version was optimized for high quality dialogue usecases. It has demonstrated strong performance compared to leading closed-source models in human evaluations. To read more about the model release, click here. Usage of this model is subject to Meta's Acceptable Use Policy. Note: this is a higher-throughput version of this model, and may have higher prices and slightly different outputs.
by meta-llama8K context$0.2/M input tkns$0.2/M output tkns93.5M tokens this week - Meta: Llama 3 70B Instruct (nitro)
Meta's latest class of model (Llama 3) launched with a variety of sizes & flavors. This 70B instruct-tuned version was optimized for high quality dialogue usecases. It has demonstrated strong performance compared to leading closed-source models in human evaluations. To read more about the model release, click here. Usage of this model is subject to Meta's Acceptable Use Policy. Note: this is a higher-throughput version of this model, and may have higher prices and slightly different outputs.
by meta-llama8K context$0.9/M input tkns$0.9/M output tkns433M tokens this week - Meta: Llama 3 8B Instruct
Meta's latest class of model (Llama 3) launched with a variety of sizes & flavors. This 8B instruct-tuned version was optimized for high quality dialogue usecases. It has demonstrated strong performance compared to leading closed-source models in human evaluations. To read more about the model release, click here. Usage of this model is subject to Meta's Acceptable Use Policy.
by meta-llama8K context$0.07/M input tkns$0.07/M output tkns2.23B tokens this week - Meta: Llama 3 70B Instruct
Meta's latest class of model (Llama 3) launched with a variety of sizes & flavors. This 70B instruct-tuned version was optimized for high quality dialogue usecases. It has demonstrated strong performance compared to leading closed-source models in human evaluations. To read more about the model release, click here. Usage of this model is subject to Meta's Acceptable Use Policy.
by meta-llama8K context$0.59/M input tkns$0.79/M output tkns4.14B tokens this week - Mistral: Mixtral 8x22B Instruct
Mistral's official instruct fine-tuned version of Mixtral 8x22B. It uses 39B active parameters out of 141B, offering unparalleled cost efficiency for its size. Its strengths include: - strong math, coding, and reasoning - large context length (64k) - fluency in English, French, Italian, German, and Spanish See benchmarks on the launch announcement here. #moe
by mistralai66K context$0.65/M input tkns$0.65/M output tkns420M tokens this week - WizardLM-2 8x22B
WizardLM-2 8x22B is Microsoft AI's most advanced Wizard model. It demonstrates highly competitive performance compared to leading proprietary models, and it consistently outperforms all existing state-of-the-art opensource models. It is an instruct finetune of Mixtral 8x22B. To read more about the model release, click here. #moe
by microsoft66K context$0.65/M input tkns$0.65/M output tkns1.75B tokens this week - WizardLM-2 7B
WizardLM-2 7B is the smaller variant of Microsoft AI's latest Wizard model. It is the fastest and achieves comparable performance with existing 10x larger opensource leading models It is a finetune of Mistral 7B Instruct, using the same technique as WizardLM-2 8x22B. To read more about the model release, click here. #moe
by microsoft32K context$0.07/M input tkns$0.07/M output tkns641M tokens this week - Toppy M 7B (nitro)
A wild 7B parameter model that merges several models using the new task_arithmetic merge method from mergekit. List of merged models: - NousResearch/Nous-Capybara-7B-V1.9 - HuggingFaceH4/zephyr-7b-beta - lemonilia/AshhLimaRP-Mistral-7B - Vulkane/120-Days-of-Sodom-LoRA-Mistral-7b - Undi95/Mistral-pippa-sharegpt-7b-qlora #merge #uncensored Note: this is a higher-throughput version of this model, and may have higher prices and slightly different outputs.
by undi954K context$0.07/M input tkns$0.07/M output tkns2.22M tokens this week - Mistral: Mixtral 8x22B (base)
Mixtral 8x22B is a large-scale language model from Mistral AI. It consists of 8 experts, each 22 billion parameters, with each token using 2 experts at a time. It was released via X. #moe
by mistralai66K context$0.9/M input tkns$0.9/M output tkns29.4M tokens this week - OpenAI: GPT-4 Turbo
The latest GPT-4 Turbo model with vision capabilities. Vision requests can now use JSON mode and function calling. Training data: up to Dec 2023. This model is updated by OpenAI to point to the latest version of GPT-4 Turbo, currently gpt-4-turbo-2024-04-09 (as of April 2024).
by openai128K context$10/M input tkns$30/M output tkns$14.45/K input imgs658M tokens this week - Google: Gemini Pro 1.5 (preview)
Google's latest multimodal model, supporting image and video in text or chat prompts. Optimized for language tasks including: - Code generation - Text generation - Text editing - Problem solving - Recommendations - Information extraction - Data extraction or generation - AI agents Usage of Gemini is subject to Google's Gemini Terms of Use. Note: Preview models are offered for testing purposes and should not be used in production apps. This model is heavily rate limited. #multimodal
by google2.8M context$2.5/M input tkns$7.5/M output tkns$2.65/K input imgs804M tokens this week - Cohere: Command R+
Command R+ is a new, 104B-parameter LLM from Cohere. It's useful for roleplay, general consumer usecases, and Retrieval Augmented Generation (RAG). It offers multilingual support for ten key languages to facilitate global business operations. See benchmarks and the launch post here. Use of this model is subject to Cohere's Acceptable Use Policy.
by cohere128K context$3/M input tkns$15/M output tkns353M tokens this week - Databricks: DBRX 132B Instruct
DBRX is a new open source large language model developed by Databricks. At 132B, it outperforms existing open source LLMs like Llama 2 70B and Mixtral-8x7b on standard industry benchmarks for language understanding, programming, math, and logic. It uses a fine-grained mixture-of-experts (MoE) architecture. 36B parameters are active on any input. It was pre-trained on 12T tokens of text and code data. Compared to other open MoE models like Mixtral-8x7B and Grok-1, DBRX is fine-grained, meaning it uses a larger number of smaller experts. See the launch announcement and benchmark results here. #moe
by databricks33K context$1.08/M input tkns$1.08/M output tkns7.08M tokens this week - Midnight Rose 70B
A merge with a complex family tree, this model was crafted for roleplaying and storytelling. Midnight Rose is a successor to Rogue Rose and Aurora Nights and improves upon them both. It wants to produce lengthy output by default and is the best creative writing merge produced so far by sophosympatheia. Descending from earlier versions of Midnight Rose and Wizard Tulu Dolphin 70B, it inherits the best qualities of each.
by sophosympatheia4K context$9/M input tkns$9/M output tkns14.9M tokens this week - Cohere: Command
Command is an instruction-following conversational model that performs language tasks with high quality, more reliably and with a longer context than our base generative models. Use of this model is subject to Cohere's Acceptable Use Policy.
by cohere4K context$1/M input tkns$2/M output tkns4.4M tokens this week - Cohere: Command R
Command-R is a 35B parameter model that performs conversational language tasks at a higher quality, more reliably, and with a longer context than previous models. It can be used for complex workflows like code generation, retrieval augmented generation (RAG), tool use, and agents. Read the launch post here. Use of this model is subject to Cohere's Acceptable Use Policy.
by cohere128K context$0.5/M input tkns$1.5/M output tkns213M tokens this week - Anthropic: Claude 3 Haiku
Claude 3 Haiku is Anthropic's fastest and most compact model for near-instant responsiveness. Quick and accurate targeted performance. See the launch announcement and benchmark results here #multimodal
by anthropic200K context$0.25/M input tkns$1.25/M output tkns$0.4/K input imgs2.78B tokens this week - Anthropic: Claude 3 Haiku (self-moderated)
This is a lower-latency version of Claude 3 Haiku, made available in collaboration with Anthropic, that is self-moderated: response moderation happens on the model's side instead of OpenRouter's. It's in beta, and may change in the future. Claude 3 Haiku is Anthropic's fastest and most compact model for near-instant responsiveness. Quick and accurate targeted performance. See the launch announcement and benchmark results here #multimodal
by anthropic200K context$0.25/M input tkns$1.25/M output tkns$0.4/K input imgs689M tokens this week - Google: Gemma 7B (nitro)
Gemma by Google is an advanced, open-source language model family, leveraging the latest in decoder-only, text-to-text technology. It offers English language capabilities across text generation tasks like question answering, summarization, and reasoning. The Gemma 7B variant is comparable in performance to leading open source models. Usage of Gemma is subject to Google's Gemma Terms of Use. Note: this is a higher-throughput version of this model, and may have higher prices and slightly different outputs.
by google8K context$0.2/M input tkns$0.2/M output tkns625K tokens this week - Mistral 7B Instruct (nitro)
A 7.3B parameter model that outperforms Llama 2 13B on all benchmarks, with optimizations for speed and context length. This is v0.2 of Mistral 7B Instruct. For v0.1, use this model. Note: this is a higher-throughput version of this model, and may have higher prices and slightly different outputs.
by mistralai33K context$0.2/M input tkns$0.2/M output tkns192M tokens this week - Mixtral 8x7B Instruct (nitro)
A pretrained generative Sparse Mixture of Experts, by Mistral AI, for chat and instruction use. Incorporates 8 experts (feed-forward networks) for a total of 47 billion parameters. Instruct model fine-tuned by Mistral. #moe Note: this is a higher-throughput version of this model, and may have higher prices and slightly different outputs.
by mistralai33K context$0.54/M input tkns$0.54/M output tkns301M tokens this week - Meta: Llama v2 70B Chat (nitro)
The flagship, 70 billion parameter language model from Meta, fine tuned for chat completions. Llama 2 is an auto-regressive language model that uses an optimized transformer architecture. The tuned versions use supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) to align to human preferences for helpfulness and safety. Note: this is a higher-throughput version of this model, and may have higher prices and slightly different outputs.
by meta-llama4K context$0.9/M input tkns$0.9/M output tkns946K tokens this week - MythoMax 13B (nitro)
One of the highest performing and most popular fine-tunes of Llama 2 13B, with rich descriptions and roleplay. #merge Note: this is a higher-throughput version of this model, and may have higher prices and slightly different outputs.
by gryphe4K context$0.2/M input tkns$0.2/M output tkns1.15B tokens this week - Anthropic: Claude 3 Opus
Claude 3 Opus is Anthropic's most powerful model for highly complex tasks. It boasts top-level performance, intelligence, fluency, and understanding. See the launch announcement and benchmark results here #multimodal
by anthropic200K context$15/M input tkns$75/M output tkns$24/K input imgs954M tokens this week - Anthropic: Claude 3 Sonnet
Claude 3 Sonnet is an ideal balance of intelligence and speed for enterprise workloads. Maximum utility at a lower price, dependable, balanced for scaled deployments. See the launch announcement and benchmark results here #multimodal
by anthropic200K context$3/M input tkns$15/M output tkns$4.8/K input imgs691M tokens this week - Anthropic: Claude 3 Opus (self-moderated)
This is a lower-latency version of Claude 3 Opus, made available in collaboration with Anthropic, that is self-moderated: response moderation happens on the model's side instead of OpenRouter's. It's in beta, and may change in the future. Claude 3 Opus is Anthropic's most powerful model for highly complex tasks. It boasts top-level performance, intelligence, fluency, and understanding. See the launch announcement and benchmark results here #multimodal
by anthropic200K context$15/M input tkns$75/M output tkns$24/K input imgs564M tokens this week - Anthropic: Claude 3 Sonnet (self-moderated)
This is a lower-latency version of Claude 3 Sonnet, made available in collaboration with Anthropic, that is self-moderated: response moderation happens on the model's side instead of OpenRouter's. It's in beta, and may change in the future. Claude 3 Sonnet is an ideal balance of intelligence and speed for enterprise workloads. Maximum utility at a lower price, dependable, balanced for scaled deployments. See the launch announcement and benchmark results here #multimodal
by anthropic200K context$3/M input tkns$15/M output tkns$4.8/K input imgs661M tokens this week - Mistral Large
This is Mistral AI's closed-source, flagship model. It's powered by a closed-source prototype and excels at reasoning, code, JSON, chat, and more. Read the launch announcement here. It is fluent in English, French, Spanish, German, and Italian, with high grammatical accuracy, and its 32K tokens context window allows precise information recall from large documents.
by mistralai32K context$8/M input tkns$24/M output tkns57.5M tokens this week - Google: Gemma 7B (free)
Gemma by Google is an advanced, open-source language model family, leveraging the latest in decoder-only, text-to-text technology. It offers English language capabilities across text generation tasks like question answering, summarization, and reasoning. The Gemma 7B variant is comparable in performance to leading open source models. Usage of Gemma is subject to Google's Gemma Terms of Use. Note: this is a free, rate-limited version of this model. Outputs may be cached. Read about rate limits here.
by google8K context$0/M input tkns$0/M output tkns11.6M tokens this week - Google: Gemma 7B
Gemma by Google is an advanced, open-source language model family, leveraging the latest in decoder-only, text-to-text technology. It offers English language capabilities across text generation tasks like question answering, summarization, and reasoning. The Gemma 7B variant is comparable in performance to leading open source models. Usage of Gemma is subject to Google's Gemma Terms of Use.
by google8K context$0.07/M input tkns$0.07/M output tkns253M tokens this week - Nous: Hermes 2 Mistral 7B DPO
This is the flagship 7B Hermes model, a Direct Preference Optimization (DPO) of Teknium/OpenHermes-2.5-Mistral-7B. It shows improvement across the board on all benchmarks tested - AGIEval, BigBench Reasoning, GPT4All, and TruthfulQA. The model prior to DPO was trained on 1,000,000 instructions/chats of GPT-4 quality or better, primarily synthetic data as well as other high quality datasets.
by nousresearch8K context$0.18/M input tkns$0.18/M output tkns400M tokens this week - Anthropic: Claude v2 (self-moderated)
This is a lower-latency version of Claude v2, made available in collaboration with Anthropic, that is self-moderated: response moderation happens on the model's side instead of OpenRouter's. It's in beta, and may change in the future. Claude 2 delivers advancements in key capabilities for enterprises—including an industry-leading 200K token context window, significant reductions in rates of model hallucination, system prompts and a new beta feature: tool use.
by anthropic200K context$8/M input tkns$24/M output tkns6.56M tokens this week - Anthropic: Claude v2.0 (self-moderated)
This is a lower-latency version of Claude v2.0, made available in collaboration with Anthropic, that is self-moderated: response moderation happens on the model's side instead of OpenRouter's. It's in beta, and may change in the future. Anthropic's flagship model. Superior performance on tasks that require complex reasoning. Supports hundreds of pages of text.
by anthropic100K context$8/M input tkns$24/M output tkns13.9M tokens this week - Anthropic: Claude v2.1 (self-moderated)
This is a lower-latency version of Claude v2.1, made available in collaboration with Anthropic, that is self-moderated: response moderation happens on the model's side instead of OpenRouter's. It's in beta, and may change in the future. Claude 2 delivers advancements in key capabilities for enterprises—including an industry-leading 200K token context window, significant reductions in rates of model hallucination, system prompts and a new beta feature: tool use.
by anthropic200K context$8/M input tkns$24/M output tkns53.4M tokens this week - Anthropic: Claude Instant v1 (self-moderated)
This is a lower-latency version of Claude Instant v1, made available in collaboration with Anthropic, that is self-moderated: response moderation happens on the model's side instead of OpenRouter's. It's in beta, and may change in the future. Anthropic's model for low-latency, high throughput text generation. Supports hundreds of pages of text.
by anthropic100K context$0.8/M input tkns$2.4/M output tkns3.19M tokens this week - OpenAI: GPT-3.5 Turbo 16k
The latest GPT-3.5 Turbo model with improved instruction following, JSON mode, reproducible outputs, parallel function calling, and more. Training data: up to Sep 2021. This version has a higher accuracy at responding in requested formats and a fix for a bug which caused a text encoding issue for non-English language function calls.
by openai16K context$0.5/M input tkns$1.5/M output tkns1.1B tokens this week - Meta: CodeLlama 70B Instruct
Code Llama is a family of large language models for code. This one is based on Llama 2 70B and provides zero-shot instruction-following ability for programming tasks.
by codellama2K context$0.81/M input tkns$0.81/M output tkns2.86M tokens this week - RWKV v5: Eagle 7B
Eagle 7B is trained on 1.1 Trillion Tokens across 100+ world languages (70% English, 15% multilang, 15% code). - Built on the RWKV-v5 architecture (a linear transformer with 10-100x+ lower inference cost) - Ranks as the world's greenest 7B model (per token) - Outperforms all 7B class models in multi-lingual benchmarks - Approaches Falcon (1.5T), LLaMA2 (2T), Mistral (>2T?) level of performance in English evals - Trade blows with MPT-7B (1T) in English evals - All while being an "Attention-Free Transformer" Eagle 7B models are provided for free, by Recursal.AI, for the beta period till end of March 2024 Find out more here rnn
by recursal10K context$0/M input tkns$0/M output tkns4.2M tokens this week - OpenAI: GPT-3.5 Turbo (older v0613)
GPT-3.5 Turbo is OpenAI's fastest model. It can understand and generate natural language or code, and is optimized for chat and traditional completion tasks. Updated by OpenAI to point to the latest version of GPT-3.5. Training data up to Sep 2021.
by openai4K context$1/M input tkns$2/M output tkns3.41M tokens this week - OpenAI: GPT-4 Turbo Preview
The latest GPT-4 model with improved instruction following, JSON mode, reproducible outputs, parallel function calling, and more. Training data: up to Dec 2023. Note: heavily rate limited by OpenAI while in preview.
by openai128K context$10/M input tkns$30/M output tkns39M tokens this week - ReMM SLERP 13B (extended)
A recreation trial of the original MythoMax-L2-B13 but with updated models. #merge Note: this is an extended-context version of this model. It may have higher prices and different outputs.
by undi956K context$1.125/M input tkns$1.125/M output tkns323M tokens this week - Nous: Hermes 2 Mixtral 8x7B SFT
Nous Hermes 2 Mixtral 8x7B SFT is the supervised finetune only version of the Nous Research model trained over the Mixtral 8x7B MoE LLM. The model was trained on over 1,000,000 entries of primarily GPT-4 generated data, as well as other high quality data from open datasets across the AI landscape, achieving state of the art performance on a variety of tasks. #moe
by nousresearch33K context$0.54/M input tkns$0.54/M output tkns9.74M tokens this week - Nous: Hermes 2 Mixtral 8x7B DPO
Nous Hermes 2 Mixtral 8x7B DPO is the new flagship Nous Research model trained over the Mixtral 8x7B MoE LLM. The model was trained on over 1,000,000 entries of primarily GPT-4 generated data, as well as other high quality data from open datasets across the AI landscape, achieving state of the art performance on a variety of tasks. #moe
by nousresearch33K context$0.27/M input tkns$0.27/M output tkns81.4M tokens this week - Mistral Tiny
This model is currently powered by Mistral-7B-v0.2, and incorporates a "better" fine-tuning than Mistral 7B, inspired by community work. It's best used for large batch processing tasks where cost is a significant factor but reasoning capabilities are not crucial.
by mistralai32K context$0.25/M input tkns$0.25/M output tkns2.41B tokens this week - Mistral Small
This model is currently powered by Mixtral-8X7B-v0.1, a sparse mixture of experts model with 12B active parameters. It has better reasoning, exhibits more capabilities, can produce and reason about code, and is multiligual, supporting English, French, German, Italian, and Spanish. #moe
by mistralai32K context$2/M input tkns$6/M output tkns20.4M tokens this week - Mistral Medium
This is Mistral AI's closed-source, medium-sided model. It's powered by a closed-source prototype and excels at reasoning, code, JSON, chat, and more. In benchmarks, it compares with many of the flagship models of other companies.
by mistralai32K context$2.7/M input tkns$8.1/M output tkns183M tokens this week - Bagel 34B v0.2
An experimental fine-tune of Yi 34b 200k using bagel. This is the version of the fine-tune before direct preference optimization (DPO) has been applied. DPO performs better on benchmarks, but this version is likely better for creative writing, roleplay, etc.
by jondurbin8K context - Chronos Hermes 13B v2
A 75/25 merge of Chronos 13b v2 and Nous Hermes Llama2 13b. This offers the imaginative writing style of Chronos while retaining coherency. Outputs are long and use exceptional prose. #merge
by austism4K context$0.13/M input tkns$0.13/M output tkns126M tokens this week - Nous: Hermes 2 Yi 34B
Nous Hermes 2 Yi 34B was trained on 1,000,000 entries of primarily GPT-4 generated data, as well as other high quality data from open datasets across the AI landscape. Nous-Hermes 2 on Yi 34B outperforms all Nous-Hermes & Open-Hermes models of the past, achieving new heights in all benchmarks for a Nous Research LLM as well as surpassing many popular finetunes.
by nousresearch4K context$0.72/M input tkns$0.72/M output tkns6.04M tokens this week - Noromaid Mixtral 8x7B Instruct
This model was trained for 8h(v1) + 8h(v2) + 12h(v3) on customized modified datasets, focusing on RP, uncensoring, and a modified version of the Alpaca prompting (that was already used in LimaRP), which should be at the same conversational level as ChatLM or Llama2-Chat without adding any additional special tokens.
by neversleep8K context$8/M input tkns$8/M output tkns174M tokens this week - Dolphin 2.6 Mixtral 8x7B 🐬
This is a 16k context fine-tune of Mixtral-8x7b. It excels in coding tasks due to extensive training with coding data and is known for its obedience, although it lacks DPO tuning. The model is uncensored and is stripped of alignment and bias. It requires an external alignment layer for ethical use. Users are cautioned to use this highly compliant model responsibly, as detailed in a blog post about uncensored models at erichartford.com/uncensored-models. #moe #uncensored
by cognitivecomputations33K context$0.5/M input tkns$0.5/M output tkns88.3M tokens this week - Google: Gemini Pro 1.0
Google's flagship text generation model. Designed to handle natural language tasks, multiturn text and code chat, and code generation. See the benchmarks and prompting guidelines from Deepmind. Usage of Gemini is subject to Google's Gemini Terms of Use.
by google92K context$0.125/M input tkns$0.375/M output tkns$2.5/K input imgs388M tokens this week - Google: Gemini Pro Vision 1.0
Google's flagship multimodal model, supporting image and video in text or chat prompts for a text or code response. See the benchmarks and prompting guidelines from Deepmind. Usage of Gemini is subject to Google's Gemini Terms of Use. #multimodal
by google46K context$0.125/M input tkns$0.375/M output tkns$2.5/K input imgs8.31M tokens this week - Mixtral 8x7B (base)
A pretrained generative Sparse Mixture of Experts, by Mistral AI. Incorporates 8 experts (feed-forward networks) for a total of 47B parameters. Base model (not fine-tuned for instructions) - see Mixtral 8x7B Instruct for an instruct-tuned model. #moe
by mistralai33K context$0.54/M input tkns$0.54/M output tkns2.97M tokens this week - Mixtral 8x7B Instruct
A pretrained generative Sparse Mixture of Experts, by Mistral AI, for chat and instruction use. Incorporates 8 experts (feed-forward networks) for a total of 47 billion parameters. Instruct model fine-tuned by Mistral. #moe
by mistralai33K context$0.24/M input tkns$0.24/M output tkns4.33B tokens this week - RWKV v5 World 3B
RWKV is an RNN (recurrent neural network) with transformer-level performance. It aims to combine the best of RNNs and transformers - great performance, fast inference, low VRAM, fast training, "infinite" context length, and free sentence embedding. RWKV-5 is trained on 100+ world languages (70% English, 15% multilang, 15% code). RWKV 3B models are provided for free, by Recursal.AI, for the beta period. More details here. #rnn
by rwkv10K context$0/M input tkns$0/M output tkns2M tokens this week - RWKV v5 3B AI Town
This is an RWKV 3B model finetuned specifically for the AI Town project. RWKV is an RNN (recurrent neural network) with transformer-level performance. It aims to combine the best of RNNs and transformers - great performance, fast inference, low VRAM, fast training, "infinite" context length, and free sentence embedding. RWKV 3B models are provided for free, by Recursal.AI, for the beta period. More details here. #rnn
by recursal10K context$0/M input tkns$0/M output tkns281K tokens this week - StripedHyena Nous 7B
This is the chat model variant of the StripedHyena series developed by Together in collaboration with Nous Research. StripedHyena uses a new architecture that competes with traditional Transformers, particularly in long-context data processing. It combines attention mechanisms with gated convolutions for improved speed, efficiency, and scaling. This model marks a significant advancement in AI architecture for sequence modeling tasks.
by togethercomputer33K context$0.18/M input tkns$0.18/M output tkns1.41M tokens this week - StripedHyena Hessian 7B (base)
This is the base model variant of the StripedHyena series, developed by Together. StripedHyena uses a new architecture that competes with traditional Transformers, particularly in long-context data processing. It combines attention mechanisms with gated convolutions for improved speed, efficiency, and scaling. This model marks an advancement in AI architecture for sequence modeling tasks.
by togethercomputer33K context$0.18/M input tkns$0.18/M output tkns186K tokens this week - Psyfighter v2 13B
The v2 of Psyfighter - a merged model created by the KoboldAI community members Jeb Carter and TwistedShadows, made possible thanks to the KoboldAI merge request service. The intent was to add medical data to supplement the model's fictional ability with more details on anatomy and mental states. This model should not be used for medical advice or therapy because of its high likelihood of pulling in fictional data. It's a merge between: - KoboldAI/LLaMA2-13B-Tiefighter - Doctor-Shotgun/cat-v1.0-13b - Doctor-Shotgun/llama-2-13b-chat-limarp-v2-merged. #merge
by koboldai4K context$1/M input tkns$1/M output tkns32M tokens this week - Yi 34B Chat
The Yi series models are large language models trained from scratch by developers at 01.AI. This version is instruct-tuned to work better for chat.
by 01-ai4K context$0.72/M input tkns$0.72/M output tkns34.5M tokens this week - Yi 34B (base)
The Yi series models are large language models trained from scratch by developers at 01.AI.
by 01-ai4K context$0.72/M input tkns$0.72/M output tkns158K tokens this week - Yi 6B (base)
The Yi series models are large language models trained from scratch by developers at 01.AI.
by 01-ai4K context$0.126/M input tkns$0.126/M output tkns402K tokens this week - Nous: Hermes 2 Vision 7B (alpha)
This vision-language model builds on innovations from the popular OpenHermes-2.5 model, by Teknium. It adds vision support, and is trained on a custom dataset enriched with function calling This project is led by qnguyen3 and teknium. #multimodal
by nousresearch4K context$10/M input tkns$10/M output tkns280 tokens this week - Cinematika 7B (alpha) (free)
This model is under development. Check the OpenRouter Discord for updates. Note: this is a free, rate-limited version of this model. Outputs may be cached. Read about rate limits here.
by openrouter8K context$0/M input tkns$0/M output tkns4.14M tokens this week - Cinematika 7B (alpha)
This model is under development. Check the OpenRouter Discord for updates.
by openrouter8K context$0.18/M input tkns$0.18/M output tkns1.09M tokens this week - Nous: Capybara 7B (free)
The Capybara series is a collection of datasets and models made by fine-tuning on data created by Nous, mostly in-house. V1.9 uses unalignment techniques for more consistent and dynamic control. It also leverages a significantly better foundation model, Mistral 7B. Note: this is a free, rate-limited version of this model. Outputs may be cached. Read about rate limits here.
by nousresearch4K context$0/M input tkns$0/M output tkns15.3M tokens this week - Nous: Capybara 7B
The Capybara series is a collection of datasets and models made by fine-tuning on data created by Nous, mostly in-house. V1.9 uses unalignment techniques for more consistent and dynamic control. It also leverages a significantly better foundation model, Mistral 7B.
by nousresearch4K context$0.18/M input tkns$0.18/M output tkns779K tokens this week - Psyfighter 13B
A #merge model based on Llama-2-13B and made possible thanks to the compute provided by the KoboldAI community. It's a merge between: - KoboldAI/LLaMA2-13B-Tiefighter - chaoyi-wu/MedLLaMA_13B - Doctor-Shotgun/llama-2-13b-chat-limarp-v2-merged. #merge
by jebcarter4K context - OpenChat 3.5 (free)
OpenChat is a library of open-source language models, fine-tuned with "C-RLFT (Conditioned Reinforcement Learning Fine-Tuning)" - a strategy inspired by offline reinforcement learning. It has been trained on mixed-quality data without preference labels. Note: this is a free, rate-limited version of this model. Outputs may be cached. Read about rate limits here.
by openchat8K context$0/M input tkns$0/M output tkns46.3M tokens this week - OpenChat 3.5
OpenChat is a library of open-source language models, fine-tuned with "C-RLFT (Conditioned Reinforcement Learning Fine-Tuning)" - a strategy inspired by offline reinforcement learning. It has been trained on mixed-quality data without preference labels.
by openchat8K context$0.07/M input tkns$0.07/M output tkns1.6B tokens this week - MythoMist 7B (free)
From the creator of MythoMax, merges a suite of models to reduce word anticipation, ministrations, and other undesirable words in ChatGPT roleplaying data. It combines Neural Chat 7B, Airoboros 7b, Toppy M 7B, Zepher 7b beta, Nous Capybara 34B, OpenHeremes 2.5, and many others. #merge Note: this is a free, rate-limited version of this model. Outputs may be cached. Read about rate limits here.
by gryphe33K context$0/M input tkns$0/M output tkns26.3M tokens this week - Noromaid 20B
A collab between IkariDev and Undi. This merge is suitable for RP, ERP, and general knowledge. #merge #uncensored
by neversleep8K context$2.25/M input tkns$2.25/M output tkns26.2M tokens this week - MythoMist 7B
From the creator of MythoMax, merges a suite of models to reduce word anticipation, ministrations, and other undesirable words in ChatGPT roleplaying data. It combines Neural Chat 7B, Airoboros 7b, Toppy M 7B, Zepher 7b beta, Nous Capybara 34B, OpenHeremes 2.5, and many others. #merge
by gryphe33K context$0.375/M input tkns$0.375/M output tkns5.72M tokens this week - Neural Chat 7B v3.1
A fine-tuned model based on mistralai/Mistral-7B-v0.1 on the open source dataset Open-Orca/SlimOrca, aligned with DPO algorithm. For more details, refer to the blog: The Practice of Supervised Fine-tuning and Direct Preference Optimization on Habana Gaudi2.
by intel4K context$5/M input tkns$5/M output tkns162K tokens this week - Anthropic: Claude v2
Claude 2 delivers advancements in key capabilities for enterprises—including an industry-leading 200K token context window, significant reductions in rates of model hallucination, system prompts and a new beta feature: tool use.
by anthropic200K context$8/M input tkns$24/M output tkns24.5M tokens this week - Anthropic: Claude v2.1
Claude 2 delivers advancements in key capabilities for enterprises—including an industry-leading 200K token context window, significant reductions in rates of model hallucination, system prompts and a new beta feature: tool use.
by anthropic200K context$8/M input tkns$24/M output tkns40.4M tokens this week - Anthropic: Claude Instant (older v1.1)
Anthropic's model for low-latency, high throughput text generation. Supports hundreds of pages of text.
by anthropic100K context$0.8/M input tkns$2.4/M output tkns5.71M tokens this week - OpenHermes 2.5 Mistral 7B
A continuation of OpenHermes 2 model, trained on additional code datasets. Potentially the most interesting finding from training on a good ratio (est. of around 7-14% of the total dataset) of code instruction was that it has boosted several non-code benchmarks, including TruthfulQA, AGIEval, and GPT4All suite. It did however reduce BigBench benchmark score, but the net gain overall is significant.
by teknium4K context$0.17/M input tkns$0.17/M output tkns159M tokens this week - LLaVA 13B
LLaVA is a large multimodal model that combines a vision encoder and Vicuna for general-purpose visual and language understanding, achieving impressive chat capabilities mimicking GPT-4 and setting a new state-of-the-art accuracy on Science QA #multimodal
by liuhaotian2K context$10/M input tkns$10/M output tkns2K tokens this week - Nous: Capybara 34B
This model is trained on the Yi-34B model for 3 epochs on the Capybara dataset. It's the first 34B Nous model and first 200K context length Nous model. Note: This endpoint currently supports 32k context.
by nousresearch33K context$0.9/M input tkns$0.9/M output tkns19.5M tokens this week - OpenAI: GPT-4 Vision
Ability to understand images, in addition to all other GPT-4 Turbo capabilties. Training data: up to Apr 2023. Note: heavily rate limited by OpenAI while in preview. #multimodal
by openai128K context$10/M input tkns$30/M output tkns$14.45/K input imgs24.9M tokens this week - lzlv 70B
A Mythomax/MLewd_13B-style merge of selected 70B models. A multi-model merge of several LLaMA2 70B finetunes for roleplaying and creative work. The goal was to create a model that combines creativity with intelligence for an enhanced experience. #merge #uncensored
by lizpreciatior4K context$0.59/M input tkns$0.79/M output tkns393M tokens this week - Toppy M 7B (free)
A wild 7B parameter model that merges several models using the new task_arithmetic merge method from mergekit. List of merged models: - NousResearch/Nous-Capybara-7B-V1.9 - HuggingFaceH4/zephyr-7b-beta - lemonilia/AshhLimaRP-Mistral-7B - Vulkane/120-Days-of-Sodom-LoRA-Mistral-7b - Undi95/Mistral-pippa-sharegpt-7b-qlora #merge #uncensored Note: this is a free, rate-limited version of this model. Outputs may be cached. Read about rate limits here.
by undi954K context$0/M input tkns$0/M output tkns84.3M tokens this week - Goliath 120B
A large LLM created by combining two fine-tuned Llama 70B models into one 120B model. Combines Xwin and Euryale. Credits to - @chargoddard for developing the framework used to merge the model - mergekit. - @Undi95 for helping with the merge ratios. #merge
by alpindale6K context$9.375/M input tkns$9.375/M output tkns73.6M tokens this week - Toppy M 7B
A wild 7B parameter model that merges several models using the new task_arithmetic merge method from mergekit. List of merged models: - NousResearch/Nous-Capybara-7B-V1.9 - HuggingFaceH4/zephyr-7b-beta - lemonilia/AshhLimaRP-Mistral-7B - Vulkane/120-Days-of-Sodom-LoRA-Mistral-7b - Undi95/Mistral-pippa-sharegpt-7b-qlora #merge #uncensored
by undi954K context$0.07/M input tkns$0.07/M output tkns70.1M tokens this week - Auto (best for prompt)
Depending on their size, subject, and complexity, your prompts will be sent to Mistral Large, Claude 3 Sonnet or GPT-4o. To see which model was used, visit Activity.
by openrouter200K context - OpenAI: GPT-3.5 Turbo 16k (older v1106)
The latest GPT-3.5 Turbo model with improved instruction following, JSON mode, reproducible outputs, parallel function calling, and more. Training data: up to Sep 2021.
by openai16K context$1/M input tkns$2/M output tkns30.9M tokens this week - OpenAI: GPT-4 Turbo (older v1106)
The latest GPT-4 model with improved instruction following, JSON mode, reproducible outputs, parallel function calling, and more. Training data: up to Apr 2023. Note: heavily rate limited by OpenAI while in preview.
by openai128K context$10/M input tkns$30/M output tkns235M tokens this week - Hugging Face: Zephyr 7B
Zephyr is a series of language models that are trained to act as helpful assistants. Zephyr-7B-β is the second model in the series, and is a fine-tuned version of mistralai/Mistral-7B-v0.1 that was trained on a mix of publicly available, synthetic datasets using Direct Preference Optimization (DPO).
by huggingfaceh44K context$0.2/M input tkns$0.2/M output tkns7.53M tokens this week - Hugging Face: Zephyr 7B (free)
Zephyr is a series of language models that are trained to act as helpful assistants. Zephyr-7B-β is the second model in the series, and is a fine-tuned version of mistralai/Mistral-7B-v0.1 that was trained on a mix of publicly available, synthetic datasets using Direct Preference Optimization (DPO). Note: this is a free, rate-limited version of this model. Outputs may be cached. Read about rate limits here.
by huggingfaceh44K context$0/M input tkns$0/M output tkns3.43M tokens this week - Google: PaLM 2 Chat 32k
PaLM 2 is a language model by Google with improved multilingual, reasoning and coding capabilities.
by google92K context$0.25/M input tkns$0.5/M output tkns18.5M tokens this week - Google: PaLM 2 Code Chat 32k
PaLM 2 fine-tuned for chatbot conversations that help with code-related questions.
by google92K context$0.25/M input tkns$0.5/M output tkns2.13M tokens this week - OpenHermes 2 Mistral 7B
Trained on 900k instructions, surpasses all previous versions of Hermes 13B and below, and matches 70B on some benchmarks. Hermes 2 has strong multiturn chat skills and system prompt capabilities.
by teknium4K context$0.18/M input tkns$0.18/M output tkns121M tokens this week - Mistral OpenOrca 7B
A fine-tune of Mistral using the OpenOrca dataset. First 7B model to beat all other models <30B.
by open-orca8K context$0.18/M input tkns$0.18/M output tkns2.41M tokens this week - Airoboros 70B
A Llama 2 70B fine-tune using synthetic data (the Airoboros dataset). Currently based on jondurbin/airoboros-l2-70b-2.2.1, but might get updated in the future.
by jondurbin4K context$0.7/M input tkns$0.9/M output tkns55.2M tokens this week - MythoMax 13B (extended)
One of the highest performing and most popular fine-tunes of Llama 2 13B, with rich descriptions and roleplay. #merge Note: this is an extended-context version of this model. It may have higher prices and different outputs.
by gryphe8K context$1.125/M input tkns$1.125/M output tkns54.2M tokens this week - Xwin 70B
Xwin-LM aims to develop and open-source alignment tech for LLMs. Our first release, built-upon on the Llama2 base models, ranked TOP-1 on AlpacaEval. Notably, it's the first to surpass GPT-4 on this benchmark. The project will be continuously updated.
by xwin-lm8K context$3.75/M input tkns$3.75/M output tkns8.98M tokens this week - Mistral 7B Instruct (free)
A 7.3B parameter model that outperforms Llama 2 13B on all benchmarks, with optimizations for speed and context length. This is v0.1 of Mistral 7B Instruct. For v0.2, use this model. Note: this is a free, rate-limited version of this model. Outputs may be cached. Read about rate limits here.
by mistralai33K context$0/M input tkns$0/M output tkns110M tokens this week - OpenAI: GPT-3.5 Turbo Instruct
This model is a variant of GPT-3.5 Turbo tuned for instructional prompts and omitting chat-related optimizations. Training data: up to Sep 2021.
by openai4K context$1.5/M input tkns$2/M output tkns4.42M tokens this week - Mistral 7B Instruct
A 7.3B parameter model that outperforms Llama 2 13B on all benchmarks, with optimizations for speed and context length. This is v0.1 of Mistral 7B Instruct. For v0.2, use this model.
by mistralai33K context$0.07/M input tkns$0.07/M output tkns3.36B tokens this week - Pygmalion: Mythalion 13B
A blend of the new Pygmalion-13b and MythoMax. #merge
by pygmalionai8K context$1.125/M input tkns$1.125/M output tkns154M tokens this week - OpenAI: GPT-3.5 Turbo 16k
This model offers four times the context length of gpt-3.5-turbo, allowing it to support approximately 20 pages of text in a single request at a higher cost. Training data: up to Sep 2021.
by openai16K context$3/M input tkns$4/M output tkns38.4M tokens this week - OpenAI: GPT-4 32k
GPT-4-32k is an extended version of GPT-4, with the same capabilities but quadrupled context length, allowing for processing up to 40 pages of text in a single pass. This is particularly beneficial for handling longer content like interacting with PDFs without an external vector database. Training data: up to Sep 2021.
by openai33K context$60/M input tkns$120/M output tkns11.8M tokens this week - OpenAI: GPT-4 32k (older v0314)
GPT-4-32k is an extended version of GPT-4, with the same capabilities but quadrupled context length, allowing for processing up to 40 pages of text in a single pass. This is particularly beneficial for handling longer content like interacting with PDFs without an external vector database. Training data: up to Sep 2021.
by openai33K context$60/M input tkns$120/M output tkns1.18M tokens this week - Meta: CodeLlama 34B Instruct
Code Llama is built upon Llama 2 and excels at filling in code, handling extensive input contexts, and folling programming instructions without prior training for various programming tasks.
by meta-llama8K context$0.72/M input tkns$0.72/M output tkns1.34M tokens this week - Phind: CodeLlama 34B v2
A fine-tune of CodeLlama-34B on an internal dataset that helps it exceed GPT-4 on some benchmarks, including HumanEval.
by phind4K context$0.72/M input tkns$0.72/M output tkns1.63M tokens this week - Nous: Hermes 13B
A state-of-the-art language model fine-tuned on over 300k instructions by Nous Research, with Teknium and Emozilla leading the fine tuning process.
by nousresearch4K context$0.18/M input tkns$0.18/M output tkns436M tokens this week - Mancer: Weaver (alpha)
An attempt to recreate Claude-style verbosity, but don't expect the same level of coherence or memory. Meant for use in roleplay/narrative situations.
by mancer8K context$3.375/M input tkns$3.375/M output tkns10.9M tokens this week - Anthropic: Claude v2.0
Anthropic's flagship model. Superior performance on tasks that require complex reasoning. Supports hundreds of pages of text.
by anthropic100K context$8/M input tkns$24/M output tkns82.8M tokens this week - Anthropic: Claude Instant v1
Anthropic's model for low-latency, high throughput text generation. Supports hundreds of pages of text.
by anthropic100K context$0.8/M input tkns$2.4/M output tkns33.1M tokens this week - Anthropic: Claude v1
Anthropic's model for low-latency, high throughput text generation. Supports hundreds of pages of text.
by anthropic100K context$8/M input tkns$24/M output tkns8.63M tokens this week - Anthropic: Claude (older v1)
Anthropic's model for low-latency, high throughput text generation. Supports hundreds of pages of text.
by anthropic100K context$8/M input tkns$24/M output tkns16.5M tokens this week - Anthropic: Claude Instant (older v1)
Anthropic's model for low-latency, high throughput text generation. Supports hundreds of pages of text.
by anthropic100K context$0.8/M input tkns$2.4/M output tkns2.62M tokens this week - ReMM SLERP 13B
A recreation trial of the original MythoMax-L2-B13 but with updated models. #merge
by undi954K context$0.27/M input tkns$0.27/M output tkns9.97M tokens this week - Google: PaLM 2 Chat
PaLM 2 is a language model by Google with improved multilingual, reasoning and coding capabilities.
by google26K context$0.25/M input tkns$0.5/M output tkns21.9M tokens this week - Google: PaLM 2 Code Chat
PaLM 2 fine-tuned for chatbot conversations that help with code-related questions.
by google20K context$0.25/M input tkns$0.5/M output tkns591K tokens this week - MythoMax 13B
One of the highest performing and most popular fine-tunes of Llama 2 13B, with rich descriptions and roleplay. #merge
by gryphe4K context$0.13/M input tkns$0.13/M output tkns2.2B tokens this week - Meta: Llama v2 13B Chat
A 13 billion parameter language model from Meta, fine tuned for chat completions
by meta-llama4K context$0.13/M input tkns$0.13/M output tkns22.5M tokens this week - Meta: Llama v2 70B Chat
The flagship, 70 billion parameter language model from Meta, fine tuned for chat completions. Llama 2 is an auto-regressive language model that uses an optimized transformer architecture. The tuned versions use supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) to align to human preferences for helpfulness and safety.
by meta-llama4K context$0.64/M input tkns$0.8/M output tkns20.7M tokens this week - OpenAI: GPT-3.5 Turbo
GPT-3.5 Turbo is OpenAI's fastest model. It can understand and generate natural language or code, and is optimized for chat and traditional completion tasks. Updated by OpenAI to point to the latest version of GPT-3.5. Training data up to Sep 2021.
by openai16K context$0.5/M input tkns$1.5/M output tkns484M tokens this week - OpenAI: GPT-3.5 Turbo (older v0301)
GPT-3.5 Turbo is OpenAI's fastest model. It can understand and generate natural language or code, and is optimized for chat and traditional completion tasks. Updated by OpenAI to point to the latest version of GPT-3.5. Training data up to Sep 2021.
by openai4K context$1/M input tkns$2/M output tkns10.8M tokens this week - OpenAI: GPT-4
OpenAI's flagship model, GPT-4 is a large-scale multimodal language model capable of solving difficult problems with greater accuracy than previous models due to its broader general knowledge and advanced reasoning capabilities. Training data: up to Sep 2021.
by openai8K context$30/M input tkns$60/M output tkns50.6M tokens this week - OpenAI: GPT-4 (older v0314)
GPT-4-0314 is the first version of GPT-4 released, with a context length of 8,192 tokens, and was supported until June 14. Training data: up to Sep 2021.
by openai8K context$30/M input tkns$60/M output tkns1.38M tokens this week