Model library
We matched open-source models to real GPUs in the catalog. Minimum VRAM, recommended GPU class and serving engine for each model.
Loading…
LLM
textLlama Community
Llama 3.3 70B
Meta · 70B · 128K ctx
Capable open-weight general-purpose LLM for chat, analysis and generation.
—H100
textApache 2.0
Qwen 2.5 32B
Alibaba · 32B · 128K ctx
Efficiency-focused open-license LLM — Apache 2.0, safe for commercial use.
—L40S
textMIT
DeepSeek R1 Distill 14B
DeepSeek · 14B · 128K ctx
Distilled reasoning model — strong step-by-step inference at low cost.
—RTX 4090
textGemma Terms
Gemma 2 27B
Google · 27B · 8K ctx
Google's open model — large-model quality on a single L40S.
—L40S
textApache 2.0
Mistral Small 24B
Mistral · 24B · 128K ctx
European-built open model — light, fast, Apache 2.0.
—RTX 4090