T1A General Intelligence
Reliable, cost efficient LLM hosting for everyday enterprise AI tasks.
- Token quota: 20–40M
- Models: Qwen3.8 9B, Mistral Small 4
Overview
T1A is designed for assistants, Q&A systems, customer support bots, and internal productivity tools.
It offers predictable performance and generous token quotas at an accessible price point.
Open weights LLMs (weights available, fully local execution).
Included Models
- Qwen3.8 9B
- Mistral Small 4
Performance
- Token quota: 20–40M
- Low latency
- Medium throughput
- Optimized for conversational workloads
Best For
- Internal assistants
- Enterprise chat
- Customer support
- Knowledge retrieval
- Low-latency workloads
- Lightweight RAG
Add Application Hosting
Small
Medium
Large
Extra Large
Enterprise
T1B High Throughput Intelligence
More performance. More tokens. Same simplicity.
- Token quota: 40–80M
- Models: Mixtral 8x22B, Mistral Medium 3.5, DeepSeek V4.1 Flash
Overview
T1B delivers higher throughput and lower latency using Mixture of Experts architectures — ideal for high
volume chat and multi tenant applications.
Open weights LLMs (weights available, fully local execution).
Included Models
- Mixtral 8x22B
- Mistral Medium 3.5
- DeepSeek V4.1 Flash
Performance
- Token quota: 40–80M
- Very low latency
- High throughput
- Optimized for concurrency
Best For
- Multi-tenant chat
- High volume chat
- Customer-facing assistants
- High concurrency
- RAG with heavy traffic
Add Application Hosting
Small
Medium
Large
Extra Large
Enterprise
T2 Multimodal Intelligence
Vision enhanced AI for documents, images, and complex inputs.
- Token quota: 10–20M
- Models: Qwen3.8 Flash-Next
Overview
T2 enables multimodal AI workloads with support for images, documents, and visual reasoning — ideal for OCR,
document processing, and multimodal RAG.
Open weights LLMs (weights available, fully local execution).
Included Models
- Qweb3.8 Flash-Next
Performance
- Token quota: 10–20M
- Medium latency
- Vision optimized GPU backend
Best For
- Document AI
- OCR
- Image-augmented RAG
- Multimodal assistants
- Compliance document processing
Add Application Hosting
Small
Medium
Large
Extra Large
Enterprise
T3 Multimodal Intelligence
High performance code generation and developer copilots.
- Token quota: 10–20M
- Models: Codestral 2, DeepSeek V4-Pro-0813 (Code-optimized variant)
Overview
T3 is optimized for code generation, refactoring, and developer automation — ideal for internal engineering teams and DevOps copilots.
Open weights LLMs (weights available, fully local execution).
Included Models
- Codestral 2
- DeepSeek V4-Pro-0813 (code-optimized variant)
Performance
- Token quota: 10–20M
- Low latency
- Optimized for long context code tasks
Best For
- Developer copilots
- Code generation
- Refactoring
- Automation pipelines
Add Application Hosting
Small
Medium
Large
Extra Large
Enterprise
T4 Premium Intelligence
Maximum accuracy, maximum control, sovereign hosting for enterprise needs.
- Token quota: 5–10M
- Models: Qwen3.8 Max, DeepSeek V4.1 Flash / V4-Pro-08-13
Overview
T4 is built for regulated industries and mission critical workloads requiring the highest accuracy, reliability, and compliance.
Open weights LLMs (weights available, fully local execution).
Included Models
- Qwen 3.8 Max
- DeepSeek V4.1 Flash / V4 Pro-0813
- By special request: Mistral Large (2026)
Performance
- Token quota: 5–10M
- Very low latency
Best For
- Banks & insurance
- Healthcare
- Public sector
- Enterprise