GLOBALATLAS AI · DEDICATED INFERENCE
Ready AI configurations
Choose a Qwen, DeepSeek or Mistral configuration with an exact checkpoint, GPU memory assessment and serving settings.
From a model to its configuration
Choose an exact model release. Each build includes an inference memory estimate, serving settings and compatible catalog hardware. Availability is checked after selection; installation is confirmed before ordering.
Qwen2.5-7B-Instruct
Private text assistants and multilingual chat
Instruction-tuned text model. Document search requires a separate retrieval/indexing service; images and audio are not supported by this checkpoint.
- GPU VRAM
- Assessment required
- Planned RAM / disk
- 32 / 100 GB
- Context
- 8192 · BF16 · 1 request
Qwen2.5-14B-Instruct
Business assistants and structured document processing
Instruction-tuned text model. Document search requires a separate retrieval/indexing service; images and audio are not supported by this checkpoint.
- GPU VRAM
- Assessment required
- Planned RAM / disk
- 64 / 150 GB
- Context
- 8192 · BF16 · 1 request
Qwen2.5-32B-Instruct
Complex instructions and a larger self-hosted language model
Instruction-tuned text model. Document search requires a separate retrieval/indexing service; images and audio are not supported by this checkpoint.
- GPU VRAM
- Assessment required
- Planned RAM / disk
- 128 / 200 GB
- Context
- 8192 · BF16 · 1 request
Mistral-7B-Instruct-v0.3
Text services and compact application backends
Instruction-tuned text model with the v3 tokenizer. Function calling needs a compatible chat template and application-side tool execution.
- GPU VRAM
- Assessment required
- Planned RAM / disk
- 32 / 100 GB
- Context
- 8192 · BF16 · 1 request
DeepSeek-R1-Distill-Qwen-14B
Reasoning experiments and domain-specific assistants
A separate DeepSeek reasoning model distilled using DeepSeek-R1 outputs on a Qwen2.5-14B base. It is not the full 671B DeepSeek-R1 model or a bundle of two models.
- GPU VRAM
- Assessment required
- Planned RAM / disk
- 64 / 150 GB
- Context
- 8192 · BF16 · 1 request
DeepSeek-R1-Distill-Qwen-32B
Larger reasoning workloads on dedicated infrastructure
A separate DeepSeek reasoning model distilled using DeepSeek-R1 outputs on a Qwen2.5-32B base. It is not the full 671B DeepSeek-R1 model or a bundle of two models.
- GPU VRAM
- Assessment required
- Planned RAM / disk
- 128 / 200 GB
- Context
- 8192 · BF16 · 1 request
Need a different model or a cluster?
Atlas will assess your requirements and pass complex configurations to our specialists.
Atlas Ai Navigator →