Now Available: Optimized DeepSeek-R1 On GMI Cloud

GMI Cloud is excited to announce that we are now hosting a dedicated DeepSeek-R1 inference endpoint, on optimized, US-based hardware.

What's DeepSeek-R1? Read our initial takeaways here.

Technical details:

• Model Provider: DeepSeek
• Type: Chat
• Parameters: 685B
• Deployment: Serverless (MaaS) or Dedicated Endpoint
• Quantization: FP16
• Context Length: The model can remember and process up to 128,000 tokens from previous inputs within a single session.

Additionally, we are offering the following distilled models:

• DeepSeek-R1-Distill-Llama-70B
• DeepSeek-R1-Distill-Qwen-32B
• DeepSeek-R1-Distill-Qwen-14B
• DeepSeek-R1-Distill-Llama-8B
• DeepSeek-R1-Distill-Qwen-7B
• DeepSeek-R1-Distill-Qwen-1.5B

Try our token-free service with unlimited usage!

Reach out for access to our dedicated endpoint here.