Price book 2026-10-22
AI model pricing
1,982 models from 39 providers: the labs' text and embedding models, open-weight models on every major host and cloud, and image, video, speech and transcription models, each priced the way its provider bills. The same price book powers the open-source SDKs. Download it as prices.json or follow every change.
Model labs
- OpenAI112 models · Text, Embeddings, Realtime voice, Image, Video, Text to speech, Speech to text
- Anthropic17 models · Text
- Google Gemini73 models · Text, Embeddings, Realtime voice, Image, Video, Text to speech, Speech to text
- xAI57 models · Text, Realtime voice, Image, Video, Speech to text
- Mistral68 models · Text, Embeddings, Text to speech, Speech to text, Document OCR
- DeepSeek10 models · Text
- Alibaba Qwen33 models · Text, Video
- Cohere20 models · Text, Embeddings, Rerank, Document OCR
- Moonshot AI11 models · Text
- Z.ai16 models · Text
- MiniMax12 models · Text, Video, Text to speech
- AI21 Labs12 models · Text
- Perplexity79 models · Text, Embeddings
Open-model hosts
- Together AI84 models · Text, Embeddings, Rerank
- Fireworks AI329 models · Text, Embeddings, Image, Speech to text, Rerank
- Groq11 models · Text, Text to speech, Speech to text
- Cerebras7 models · Text
- SambaNova8 models · Text
- DeepInfra136 models · Text
- Novita AI135 models · Text, Embeddings, Rerank
- Nebius62 models · Text, Embeddings
- Hyperbolic16 models · Text
- Baseten32 models · Text
- Lambda20 models · Text
- Replicate40 models · Text
- Cloudflare Workers AI32 models · Text, Speech to text
Clouds
Image, video and audio
- fal84 models · Text, Image, Video
- Stability AI23 models · Image
- Black Forest Labs8 models · Image
- Runway14 models · Image, Video, Text to speech
- ElevenLabs5 models · Text to speech, Speech to text
- Deepgram38 models · Speech to text
- AssemblyAI2 models · Speech to text
- Voyage AI31 models · Embeddings, Rerank
- Jina AI1 model · Rerank
Token rates in USD per million tokens; other models per image, second, minute, character, query or page as noted. Sources: each provider's pricing page and LiteLLM's MIT-licensed price file, with Veqen's verified rules for caching, tiers and residency.