kacper-daftcode / vLLM-Moet
Frontier MoE models on RTX PRO 6000 / RTX 5090 (sm_120): DeepSeek-V4.1-Flash on 4x RTX PRO 6000, as a GPU+CPU hybrid on 4x RTX 5090, and 2-bit experts on vLLM v0.24.0
- Repository stars
- 568All-time #57,100
- Fork
- 50
- Last push
- updated 1 days ago
- Repository created
- since 2026
About
Frontier MoE models on RTX PRO 6000 / RTX 5090 (sm_120): DeepSeek-V4.1-Flash on 4x RTX PRO 6000, as a GPU+CPU hybrid on 4x RTX 5090, and 2-bit experts on vLLM v0.24.0Summarised by AI from the README; the original prevails
Author description: Frontier MoE models on RTX PRO 6000 / RTX 5090 (sm_120): DeepSeek-V4.1-Flash on 4x RTX PRO 6000, as a GPU+CPU hybrid on 4x RTX 5090, and 2-bit experts on vLLM v0.24.0
Basic listing based on official public information; functionality has not been independently evaluated.
- Pricing
- Pricing not verifiedCheck current official software, hosting, model and infrastructure pricing.