AI API providers

Alibaba Cloud's Model Studio serves the Qwen family of large language, reasoning, vision-language and translation models via a pay-as-you-go API.

Amazon's Nova family of first-party multimodal, speech and creative-generation models, served on-demand through Amazon Bedrock.

AI safety company behind the Claude model family (Fable, Opus, Sonnet, Haiku tiers), accessed via the Claude API. Anthropic does not offer its own embedding models; Claude embeddings are provided through partner Voyage AI.

Cohere builds enterprise-focused Command language models, Embed text/multimodal embeddings, and Rerank models, sold via API and dedicated Model Vault deployments.

Chinese AI lab offering a low-cost, long-context API for its DeepSeek V-series chat and reasoning models.

Google's Gemini Developer API provides pay-as-you-go access to the Gemini family of text, multimodal, and embedding models.

Meta develops the open-weight Llama model family. Its own Llama API (llama.developer.meta.com) is currently waitlist-only with no published pricing; Llama models are instead served by third-party inference providers and aggregators such as OpenRouter.

French AI lab offering a pay-as-you-go API spanning the Mistral, Magistral, Devstral, Codestral, and Voxtral model families.

AI research and deployment company behind the GPT model family, o-series reasoning models, and text-embedding models, accessed via the OpenAI API.

Voyage AI (a MongoDB company) builds specialized text and multimodal embedding models and rerankers optimized for retrieval quality.

xAI builds the Grok family of frontier language, image, video and voice models, offered via a developer API.