This has a context length of up to 8,000 tokens, making it well-suited for a wide range of advanced general language, as well as support within Retrieval Augmented Generation (RAG

Claude Haiku 3 (APAC)

Claude Haiku 3 (EU)

Claude Haiku 3 (US)

Claude 3 Sonnet by Anthropic strikes the ideal balance between intelligence and speed—particularly for enterprise workloads. It offers maximum utility at a lower price than competitors, and is engineered to be the dependable, high-endurance workhorse for scaled AI deployments. Claude 3 Sonnet can process images and return text outputs, and features a 200K context window. (APAC)

Claude Sonnet 4.6 (JP)

Claude Haiku 4.5 is the fastest, most cost-efficient model in the Claude family, delivering enhanced performance for coding, tool use, and reasoning tasks.

Claude Haiku 4.5 is the fastest, most cost-efficient model in the Claude family, delivering enhanced performance for coding, tool use, and reasoning tasks. EU regional endpoint.

Claude Haiku 4.5 is the fastest, most cost-efficient model in the Claude family, delivering enhanced performance for coding, tool use, and reasoning tasks. Global endpoint with dynamic routing.

Claude Haiku 4.5 is the fastest, most cost-efficient model in the Claude family, delivering enhanced performance for coding, tool use, and reasoning tasks. Japan regional endpoint.

Claude Haiku 4.5 is the fastest, most cost-efficient model in the Claude family, delivering enhanced performance for coding, tool use, and reasoning tasks. US regional endpoint.

Claude Haiku 4.5 is the fastest, most cost-efficient model in the Claude family. Australia regional endpoint.

Claude Opus 4.1 advances coding performance to 74.5% on SWE-bench Verified with improved in-depth research and data analysis skills. Notable performance gains in multi-file code refactoring.

Claude Opus 4.5 is Anthropic's most intelligent model, offering the highest level of performance on complex tasks requiring deep reasoning, nuanced understanding, and sophisticated analysis.

Claude Opus 4.5 is Anthropic's most intelligent model, offering the highest level of performance on complex tasks requiring deep reasoning, nuanced understanding, and sophisticated analysis.

Claude Opus 4.1 is an enhanced version of Opus 4 with improved capabilities. US regional endpoint.

Claude Opus 4.5 is Anthropic's most intelligent model with breakthrough capabilities. EU regional endpoint.

Claude Opus 4.5 is Anthropic's most intelligent model with breakthrough capabilities. US regional endpoint.

Claude Opus 4 is the world's best coding model, with sustained performance on complex, long-running tasks and agent workflows.

Claude Sonnet 4 offers excellent performance for enterprise workloads. APAC regional endpoint.

Claude Sonnet 4 offers excellent performance for enterprise workloads. Global endpoint with dynamic routing.

Claude Sonnet 4 offers excellent performance for enterprise workloads with improved instruction following. US regional endpoint.

Claude Sonnet 4.5 is a significant upgrade to Claude Sonnet 4, delivering superior coding and reasoning while responding more precisely to your instructions.

Claude Sonnet 4.5 is a significant upgrade to Claude Sonnet 4, delivering superior coding and reasoning while responding more precisely to your instructions.

Claude Sonnet 4 is Anthropic's next-generation AI model with enhanced reasoning capabilities, improved performance on complex tasks, and more nuanced understanding. This model represents a significant advancement in conversational AI with superior context handling and response quality.

Claude Sonnet 4.5 delivers superior coding and reasoning. Australia regional endpoint.

Claude Sonnet 4.5 delivers superior coding and reasoning. Japan regional endpoint.

Claude Sonnet 4.5 delivers superior coding and reasoning while responding more precisely to instructions. US regional endpoint.

Claude Sonnet 4 is a significant upgrade to Claude Sonnet 3.7, delivering superior coding and reasoning while responding more precisely to your instructions.

Meta has developed and publicly released the Llama 2 family of large language models (LLMs), a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion parameters. Our fine-tuned LLMs, called Llama-2-Chat, are optimized for dialogue use cases. Llama-2-Chat models outperform open-source chat models on most benchmarks we tested, and in our human evaluations for helpfulness and safety, are on par with some popular closed-source models like ChatGPT and PaLM. We provide a detailed description of our approach to fine-tuning and safety improvements of Llama-2-Chat in order to enable the community to build on our work and contribute to the responsible development of LLMs

Meta has developed and publicly released the Llama 2 family of large language models (LLMs), a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion parameters. Our fine-tuned LLMs, called Llama-2-Chat, are optimized for dialogue use cases. Llama-2-Chat models outperform open-source chat models on most benchmarks we tested, and in our human evaluations for helpfulness and safety, are on par with some popular closed-source models like ChatGPT and PaLM. We provide a detailed description of our approach to fine-tuning and safety improvements of Llama-2-Chat in order to enable the community to build on our work and contribute to the responsible development of LLMs

Meta has developed and publicly released the Llama 2 family of large language models (LLMs), a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion parameters. Our fine-tuned LLMs, called Llama-2-Chat, are optimized for dialogue use cases. Llama-2-Chat models outperform open-source chat models on most benchmarks we tested, and in our human evaluations for helpfulness and safety, are on par with some popular closed-source models like ChatGPT and PaLM. We provide a detailed description of our approach to fine-tuning and safety improvements of Llama-2-Chat in order to enable the community to build on our work and contribute to the responsible development of LLMs

Meta has developed and publicly released the Llama 2 family of large language models (LLMs), a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion parameters. Our fine-tuned LLMs, called Llama-2-Chat, are optimized for dialogue use cases. Llama-2-Chat models outperform open-source chat models on most benchmarks we tested, and in our human evaluations for helpfulness and safety, are on par with some popular closed-source models like ChatGPT and PaLM. We provide a detailed description of our approach to fine-tuning and safety improvements of Llama-2-Chat in order to enable the community to build on our work and contribute to the responsible development of LLMs

Meta has developed and publicly released the Llama 2 family of large language models (LLMs), a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion parameters. Our fine-tuned LLMs, called Llama-2-Chat, are optimized for dialogue use cases. Llama-2-Chat models outperform open-source chat models on most benchmarks we tested, and in our human evaluations for helpfulness and safety, are on par with some popular closed-source models like ChatGPT and PaLM. We provide a detailed description of our approach to fine-tuning and safety improvements of Llama-2-Chat in order to enable the community to build on our work and contribute to the responsible development of LLMs

Meta has developed and publicly released the Llama 2 family of large language models (LLMs), a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion parameters. Our fine-tuned LLMs, called Llama-2-Chat, are optimized for dialogue use cases. Llama-2-Chat models outperform open-source chat models on most benchmarks we tested, and in our human evaluations for helpfulness and safety, are on par with some popular closed-source models like ChatGPT and PaLM. We provide a detailed description of our approach to fine-tuning and safety improvements of Llama-2-Chat in order to enable the community to build on our work and contribute to the responsible development of LLMs

Meta developed and released the Meta Llama 3 family of large language models (LLMs), a collection of pretrained and instruction tuned generative text models in 8 and 70B sizes. The Llama 3 instruction tuned models are optimized for dialogue use cases and outperform many of the available open source chat models on common industry benchmarks. Further, in developing these models, we took great care to optimize helpfulness and safety.

Meta developed and released the Meta Llama 3 family of large language models (LLMs), a collection of pretrained and instruction tuned generative text models in 8 and 70B sizes. The Llama 3 instruction tuned models are optimized for dialogue use cases and outperform many of the available open source chat models on common industry benchmarks. Further, in developing these models, we took great care to optimize helpfulness and safety.

The Meta Llama 3.1 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction tuned generative models in 8B, 70B and 405B sizes (text in/text out). The Llama 3.1 instruction tuned text only models (8B, 70B, 405B) are optimized for multilingual dialogue use cases and outperform many of the available open source and closed chat models on common industry benchmarks.

o3-mini is our newest small reasoning model, providing high intelligence at the same cost and latency targets of o1-mini. o3-mini supports key developer features, like Structured Outputs, function calling, and Batch API.

Aya Expanse 8B

Aya Expanse 32B

Aya Vision 8B

Aya Vision 32B

DeepSeek R1 Distill Llama 70B model optimized for fast inference on Cerebras hardware. Supports up to 65,536 tokens context length.

OpenAI GPT OSS 120B model optimized for fast inference on Cerebras hardware. Supports up to 8,192 tokens context length.

Llama 3.3 70B model optimized for fast inference on Cerebras hardware. Supports up to 128,000 tokens context length.

Llama 3.1 8B model optimized for fast inference on Cerebras hardware. Supports up to 8,192 tokens context length.

Qwen 3 32B model optimized for fast inference on Cerebras hardware. Supports up to 16,382 tokens context length.

Qwen 3 235B Instruct 2507 model optimized for fast inference on Cerebras hardware. Only non-thinking mode.

GPT-4o (“o” for “omni”) is our most advanced model. It is multimodal (accepting text or image inputs and outputting text), and it has the same high intelligence as GPT-4 Turbo but is much more efficient—it generates text 2x faster and is 50% cheaper. Additionally, GPT-4o has the best vision and performance across non-English languages.

Updated version of Claude 2 with improved accuracy

Claude Haiku 4.5 (latest)

Claude Haiku 4.5 is the fastest, most cost-efficient model in the Claude family, delivering enhanced performance for coding, tool use, and reasoning tasks.

Claude Opus 4 (latest)

Claude Opus 4.1 (latest)

Claude Opus 4.5 (latest)

Claude Opus 4.5 is Anthropic's most intelligent model, offering the highest level of performance on complex tasks requiring deep reasoning, nuanced understanding, and sophisticated analysis.

Claude Opus 4.6

Claude Opus 4.6

Claude Opus 4 is the world's best coding model, with sustained performance on complex, long-running tasks and agent workflows.

Claude Opus 4.1 advances coding performance to 74.5% on SWE-bench Verified with improved in-depth research and data analysis skills. Notable performance gains in multi-file code refactoring.

Claude Sonnet 4 (latest)

Claude Sonnet 4.5 (latest)

Claude Sonnet 4.5 is a significant upgrade to Claude Sonnet 4, delivering superior coding and reasoning while responding more precisely to your instructions.

Claude Sonnet 4.6

Claude Sonnet 4.6

Claude Sonnet 4 is a significant upgrade to Claude Sonnet 3.7, delivering superior coding and reasoning while responding more precisely to your instructions.

Codestral

CogView-4 is a bilingual (Chinese/English) text-to-image generation model. Supports any-length text prompts, variable resolutions, and text-within-image generation. Ranked first in DPG-Bench benchmark for text-to-image models.

Command A is our most performant model to date, excelling at tool use, agents, retrieval augmented generation (RAG), and multilingual use cases. Command A has a context length of 256K, only requires two GPUs to run, and has 150% higher throughput compared to Command R+ 08-2024.

Command A Reasoning is Cohere's first reasoning model, able to "think" before generating an output in a way that allows it to perform well in certain kinds of nuanced problem-solving and agent-based tasks in 23 languages.

Command A Translate is Cohere's state of the art machine translation model, excelling at a variety of translation tasks on 23 languages: English, French, Spanish, Italian, German, Portuguese, Japanese, Korean, Chinese, Arabic, Russian, Polish, Turkish, Vietnamese, Dutch, Czech, Indonesian, Ukrainian, Romanian, Greek, Hindi, Hebrew, Persian.

Command A Vision is our first model capable of processing images, excelling in enterprise use cases such as analyzing charts, graphs, and diagrams, table understanding, OCR, document Q&A, and object detection. It officially supports English, Portuguese, Italian, French, German, and Spanish.

command-r-08-2024 is an update of the Command R model, delivered in August 2024. An instruction-following conversational model that performs language tasks at a higher quality, more reliably, and with a longer context than previous models.

command-r-plus-08-2024 is an update of the Command R+ model, delivered in August 2024. An instruction-following conversational model that performs language tasks at a higher quality, more reliably, and with a longer context. Best suited for complex RAG workflows and multi-step tool use.

command-r7b-12-2024 is a small, fast update delivered in December 2024. It excels at RAG, tool use, agents, and similar tasks requiring complex reasoning and multiple steps.

Command R7B Arabic

Contextual AI Rerank v1 instruction-following reranker. First generation instruction-following reranking model.

Contextual AI Rerank v2 multilingual instruction-following reranker. Best-in-class performance with support for custom instructions.

Contextual AI Rerank v2 multilingual mini instruction-following reranker. Efficient smaller model with excellent performance.

DALL·E 2 can create original, realistic images and art from a text description. It can combine concepts, attributes, and styles.

Performance on par with OpenAI-o1

DeepSeek-V3 0324 achieves a significant breakthrough in inference speed over previous models. It tops the leaderboard among open-source models and rivals the most advanced closed-source models globally.

DeepSeek-V3.1 is the latest version of the hybrid (thinking and non-thinking) model from DeepSeek. It tops the leaderboard among open-source models and rivals the most advanced closed-source models globally.

DeepSeek V3.1

DeepSeek-V3, a powerful Mixture-of-Experts language model with 671B parameters, excelling at general chat, coding, and complex reasoning tasks.

DeepSeek-R1, a reasoning model with chain-of-thought capabilities, excelling at complex reasoning, mathematics, and coding tasks with step-by-step problem solving.

Devstral 2

Devstral Medium

Devstral 2

Devstral Small 2505

Devstral Small

Ultra-fast model optimized for real-time use (~75ms†) - English only

Ultra-fast model optimized for real-time use (~75ms†)

Our most lifelike model with rich emotional expression

Our high quality, low latency model in 32 languages. Best for developer use cases where speed matters and you need non-English languages.

Light English embedding model, version 3.0, with 384-dimensional embeddings

English embedding model, version 3.0, with 1024-dimensional embeddings

Light multilingual embedding model, version 3.0, with 384-dimensional embeddings

Multilingual embedding model, version 3.0, with 1024-dimensional embeddings

English embedding model, version 3.0, with 1024-dimensional embeddings

Claude Opus 4.6 (EU)

Claude Sonnet 4.6 (EU)

FLUX.1, a 12B parameters text-to-image model with outstanding aesthetics.

FLUX.1 [pro] new is an accelerated version of FLUX.1 [pro], maintaining professional-grade image quality while delivering significantly faster generation speeds.

FLUX.1 [schnell] is a 12 billion parameter flow transformer that generates high-quality images from text in 1 to 4 steps, suitable for personal and commercial use.

State-of-the-art image generation and editing model

Gemini 1.5 Pro Experiment 0801 delivers enhanced performance with experimental features.

Gemini 1.5 Pro Experiment 0827 delivers enhanced performance with experimental features.

Gemini 2.0 Flash - Fast multimodal model with enhanced reasoning capabilities and broad input/output support.

Multimodal understanding. Realtime streaming. Native tool use.

Multimodal understanding. Realtime streaming. Native tool use.

Gemini 2.0 Flash delivers enhanced performance, doubling the speed of 1.5 Pro while supporting multimodal inputs and outputs, including text, images, and multilingual text-to-speech. It also enables seamless tool integration for advanced functionality like code execution and third-party functions.

Gemini 2.0 Flash Lite

Gemini 2.0 Flash Lite - Ultra-fast, lightweight model optimized for high-volume, low-latency tasks. Ideal for simple operations at scale.

Gemini 2.0 Flash Lite - Ultra-fast, lightweight model optimized for high-volume, low-latency tasks. Ideal for simple operations at scale.

Gemini 2.5 Flash - Fast and efficient model for high-volume tasks with enhanced capabilities. Generally available version.

Gemini 2.5 Flash - Fast and efficient model for high-volume tasks with enhanced capabilities. Generally available version.

Gemini 2.5 Flash-Lite - Our most cost-efficient and fastest 2.5 model yet. Optimized for high-volume, low-latency tasks.

Gemini 2.5 Flash-Lite - Our most cost-efficient and fastest 2.5 model yet. Optimized for high-volume, low-latency tasks.

Gemini 2.5 Flash Lite Preview (09/2025) - Ultra-fast lightweight model optimized for speed with comprehensive multimodal support.

Gemini 2.5 Flash Lite Preview 09-25

Gemini 2.5 Pro - Advanced reasoning and capabilities for complex tasks. Generally available version.

Gemini 2.5 Pro - Advanced reasoning and capabilities for complex tasks. Generally available version.

Gemini 2.0 Flash delivers enhanced performance, doubling the speed of 1.5 Pro while supporting multimodal inputs and outputs, including text, images, and multilingual text-to-speech. It also enables seamless tool integration for advanced functionality like code execution and third-party functions.

Our most intelligent model built for speed, combining frontier intelligence with superior search and grounding.

Our most intelligent model built for speed, combining frontier intelligence with superior search and grounding.

Gemini 3 Pro Preview - Advanced reasoning and capabilities for complex tasks. Preview version.

Gemini 3 Pro Preview - Advanced reasoning and capabilities for complex tasks. Preview version.

Gemini 3.1 Flash Lite Preview - Most cost-efficient Gemini model, optimized for high-volume, cost-sensitive LLM traffic with significant quality increase over previous Flash-Lite models.

Gemini 3.1 Flash Lite Preview - Most cost-efficient Gemini model, optimized for high-volume, cost-sensitive LLM traffic with significant quality increase over previous Flash-Lite models.

Gemini 3.1 Pro Preview

Gemini 3.1 Pro Preview

Gemini 3.1 Pro Preview Custom Tools

Latest Gemini embedding model with enhanced capabilities. Recommended replacement for text-embedding-004.

Gemini Embedding 001

Gemini Flash Latest

Gemini Flash-Lite Latest

Gemini Live 2.5 Flash Preview Native Audio

Gemma 2 9B is a lightweight, state-of-the-art model from Google, offering strong performance for a variety of text generation tasks with an 8k context window.

GLM-4.5 is a 355B parameter Mixture-of-Experts model with 32B active parameters. Features 128K context window, hybrid reasoning modes (Thinking/Non-Thinking), and generation speeds over 100 tokens per second. First among open-source models globally.

GLM-4.5-Air

GLM-4.5-Flash

GLM-4.5V is a 106B parameter multimodal model with 12B activated parameters. Supports vision, image, video, text, and file inputs with SOTA performance across 42 visual multimodal benchmarks. Includes "Thinking Mode" for balancing speed and reasoning depth.

GLM-4.6 is an advanced AI model with 200K context window and 128K maximum output tokens. Features enhanced reasoning, superior coding capabilities, and improved multilingual translation. Performance comparable to Claude Sonnet 4/4.6 with over 30% more token efficiency.

GLM-4.6V

GLM-4.7 is a 355B parameter Mixture-of-Experts model with 32B active parameters. Features 200K context window, 128K maximum output tokens, and enhanced agentic coding capabilities. Delivers comprehensive upgrades in general conversation, reasoning, and agent capabilities with 38% improvement over GLM-4.6 on HLE benchmarks.

GLM-4.7-Flash is a free, lightweight version of GLM-4.7 optimized for speed and cost efficiency. Provides an entry point for developers to experiment with Z.AI capabilities at no cost.

GLM-5 is a 744B parameter Mixture-of-Experts model with 40B active parameters. Features 200K context window, 128K maximum output tokens, and is designed for agentic engineering. Trained on 28.5T tokens with enhanced coding, reasoning, and agent capabilities.

GLM-5-Turbo

Claude Opus 4.6 (Global)

Claude Sonnet 4.5 (Global)

Claude Sonnet 4.6 (Global)

Most capable GPT-3.5 model and optimized for chat at 1/10th the cost of text-davinci-003

Same capabilities as the standard gpt-3.5-turbo model but with 4 times the context.

This model will also have various improvements including 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

Similar capabilities as text-davinci-003 but compatible with legacy Completions endpoint and not Chat Completions.

This model completes tasks like code generation more thoroughly than the previous preview model and is intended to reduce cases of “laziness” where the model doesn’t complete a task. The new model also includes the fix for the bug impacting non-English UTF-8 generations.

More capable than any GPT-3.5 model, able to do more complex tasks, and optimized for chat. Will be updated with our latest model iteration 2 weeks after it is released.

With 128k context, fresher knowledge and the broadest set of capabilities, GPT-4 Turbo is more powerful than GPT-4 and offered at a lower price. Will always point to GPT-4 Turbo preview model

With 128k context, fresher knowledge and the broadest set of capabilities, GPT-4 Turbo is more powerful than GPT-4 and offered at a lower price. Will always point to GPT-4 Turbo preview model

With 128k context, fresher knowledge and the broadest set of capabilities, GPT-4 Turbo is more powerful than GPT-4 and offered at a lower price.

GPT-4 with Vision, sometimes referred to as GPT-4V or gpt-4-vision-preview in the API, allows the model to take in images and answer questions about them

Outperform GPT-4o and GPT-4o mini across the board, with major gains in coding and instruction following. They also have larger context windows—supporting up to 1 million tokens of context—and are able to better use that context with improved long-context comprehension

Outperform GPT-4o and GPT-4o mini across the board, with major gains in coding and instruction following. They also have larger context windows—supporting up to 1 million tokens of context—and are able to better use that context with improved long-context comprehension

Outperform GPT-4o and GPT-4o mini across the board, with major gains in coding and instruction following. They also have larger context windows—supporting up to 1 million tokens of context—and are able to better use that context with improved long-context comprehension

Outperform GPT-4o and GPT-4o mini across the board, with major gains in coding and instruction following. They also have larger context windows—supporting up to 1 million tokens of context—and are able to better use that context with improved long-context comprehension

Outperform GPT-4o and GPT-4o mini across the board, with major gains in coding and instruction following. They also have larger context windows—supporting up to 1 million tokens of context—and are able to better use that context with improved long-context comprehension

Outperform GPT-4o and GPT-4o mini across the board, with major gains in coding and instruction following. They also have larger context windows—supporting up to 1 million tokens of context—and are able to better use that context with improved long-context comprehension

Outperform GPT-4o and GPT-4o mini across the board, with major gains in coding and instruction following. They also have larger context windows—supporting up to 1 million tokens of context—and are able to better use that context with improved long-context comprehension

Outperform GPT-4o and GPT-4o mini across the board, with major gains in coding and instruction following. They also have larger context windows—supporting up to 1 million tokens of context—and are able to better use that context with improved long-context comprehension

Outperform GPT-4o and GPT-4o mini across the board, with major gains in coding and instruction following. They also have larger context windows—supporting up to 1 million tokens of context—and are able to better use that context with improved long-context comprehension

GPT-4o (“o” for “omni”) is our most advanced model. It is multimodal (accepting text or image inputs and outputting text), and it has the same high intelligence as GPT-4 Turbo but is much more efficient—it generates text 2x faster and is 50% cheaper. Additionally, GPT-4o has the best vision and performance across non-English languages.

Groundbreaking multimodal model integrates text, vision, and audio capabilities, setting a new standard for generative and conversational AI experiences

GPT-4o (“o” for “omni”) is our most advanced model. It is multimodal (accepting text or image inputs and outputting text), and it has the same high intelligence as GPT-4 Turbo but is much more efficient—it generates text 2x faster and is 50% cheaper. Additionally, GPT-4o has the best vision and performance across non-English languages.

GPT-4o (“o” for “omni”) is our most advanced model. It is multimodal (accepting text or image inputs and outputting text), and it has the same high intelligence as GPT-4 Turbo but is much more efficient—it generates text 2x faster and is 50% cheaper. Additionally, GPT-4o has the best vision and performance across non-English languages.

GPT-4o (2024-11-20)

GPT-4o mini surpasses GPT-3.5 Turbo and other small models on academic benchmarks across both textual intelligence and multimodal reasoning, and supports the same range of languages as GPT-4o

GPT-4o mini is our most cost-efficient small model that’s smarter and cheaper than GPT-3.5 Turbo, and has vision capabilities. The model has 128K context and an October 2023 knowledge cutoff.