> ## Documentation Index > Fetch the complete documentation index at: https://agno-v2-codex-docs-audit-20260719-0149.mintlify.site/llms.txt > Use this file to discover all available pages before exploring further. # Model Providers > Setup guides for model providers and API adapters available in Agno. Browse the provider guides available in this documentation. Capabilities vary by provider endpoint and model ID. ## Provider APIs Call Anthropic's Messages API with Claude models. Call Cohere's Chat API with Command models. Run Qwen models through Alibaba Cloud DashScope. Send chat and reasoning requests to DeepSeek. Generate content with Google's Gemini API. Google's stateful Interactions API and managed agents. Inception Labs Mercury diffusion models. Meta Llama native and OpenAI-compatible adapters. Send OpenAI-compatible requests to MiniMax. Call Mistral's Chat Completions API. Use OpenAI's Chat Completions interface. Use OpenAI's Responses interface and built-in tools. Run Perplexity Sonar models and retain URL citations. Vercel v0 models for web application generation. Run Grok models through xAI. Send OpenAI-compatible requests to Xiaomi MiMo. ## Local Runtimes Local GGUF model inference with llama.cpp. Connect to LM Studio's local model server. Ollama Chat and Responses API adapters. vLLM high-throughput inference. ## Cloud Platforms Call models through the Amazon Bedrock Converse API. Anthropic Claude models via AWS Bedrock. Connect to Azure AI Foundry inference endpoints. Anthropic Claude models via Azure AI Foundry. Connect to an Azure OpenAI deployment. Anthropic Claude models via Google Vertex AI. Run foundation models through IBM watsonx.ai. ## Hosted Inference and Gateways Route OpenAI-compatible requests through AI/ML API. Call Cerebras models with the native SDK. Call Cerebras through its OpenAI-compatible endpoint. Cloudflare AI Gateway unified API. Route OpenAI-compatible requests through CometAPI. Send OpenAI-compatible requests to DeepInfra. Send OpenAI-compatible requests to Fireworks AI. Run hosted models through GroqCloud. Call models through Hugging Face Inference Providers. Route model requests through LangDB. LiteLLM unified model interface. Connect to a LiteLLM OpenAI-compatible endpoint. Run hosted models through Nebius Token Factory. Send OpenAI-compatible requests to Neosantara. Route model requests through Nexus. Connect to NVIDIA-hosted model endpoints. OpenRouter Chat Completions and Responses API adapters. Route requests through the Portkey AI Gateway. Route model requests through Requesty. Run hosted models through SambaNova Cloud. Send OpenAI-compatible requests to SiliconFlow. Run hosted models through Together AI. Connect to a governed OpenAI-compatible endpoint. ## Model String Key Catalog These are the canonical provider keys registered by the v2.7.4 model-string resolver. Similar keys can select different API adapters. | Category | Canonical keys | | ----------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Provider APIs | `anthropic`, `cohere`, `dashscope`, `deepseek`, `google`, `google-interactions`, `inception`, `internlm`, `llama-openai`, `meta`, `minimax`, `mistral`, `moonshot`, `open-responses`, `openai`, `openai-chat`, `openai-responses`, `perplexity`, `vercel`, `xai`, `xiaomi` | | Local runtimes | `llama-cpp`, `lmstudio`, `ollama`, `ollama-responses`, `vllm` | | Cloud platforms | `aws-bedrock`, `aws-claude`, `azure-ai-foundry`, `azure-foundry-claude`, `azure-openai`, `ibm`, `vertexai-claude` | | Hosted inference and gateways | `aimlapi`, `cerebras`, `cerebras-openai`, `cloudflare`, `cometapi`, `deepinfra`, `fireworks`, `groq`, `huggingface`, `langdb`, `litellm`, `litellm-openai`, `n1n`, `nebius`, `neosantara`, `nexus`, `nvidia`, `openrouter`, `openrouter-responses`, `portkey`, `requesty`, `sambanova`, `siliconflow`, `together`, `tokenlab`, `tuning-engines` | Use canonical keys in new configuration. See [Model as String](/models/model-as-string) for parsing rules and API-variant examples. ## SDK Adapters Without Dedicated Guides The SDK also exports these provider adapters, but this documentation does not yet include setup guides for them: | Adapter | Import | Related documentation | | -------- | ------------------------------------------- | ------------------------------------------------------- | | InternLM | `from agno.models.internlm import InternLM` | [InternLM reference](/reference/models/internlm) | | Moonshot | `from agno.models.moonshot import MoonShot` | [Moonshot examples](/examples/models/moonshot/overview) | | N1N | `from agno.models.n1n import N1N` | [N1N reference](/reference/models/n1n) | In v2.7.4, the `InternLM` default `base_url` already ends in `/chat/completions`, and the inherited OpenAI client appends that route again. The `internlm:model_id` string therefore sends requests to a duplicated path. Until the adapter is fixed, instantiate the class with `base_url="https://internlm-chat.intern-ai.org.cn/puyu/api/v1"` instead of using the string form. Feature support still depends on the selected model ID and endpoint. Verify both before deployment. ## Developer Resources * [OpenAI-compatible and Open Responses adapters](/models/providers/openai-like) * [Compatibility overview](/models/compatibility) * [Base model reference](/reference/models/model)