🛠️ DeepSeek-Coder-V2-Lite Under 5GB: 65 TPS 100% In-VRAM on 6GB GPU
Summary
DeepSeek-Coder-V2-Lite Under 5GB is a GGUF model optimized to run entirely in VRAM on a 6GB GPU, achieving 65 tokens per second. It provides a lightweight local coding model option for constrained hardware.
Why it’s interesting
It runs fully in VRAM on a modest 6GB GPU with high throughput under 5GB size.
Target user
Developers and users with low-spec GPUs running local coding models
Source metrics: Points 3 · Comments 0
HN discussion · Project
Source: #HackerNews / Show HN
Summary
DeepSeek-Coder-V2-Lite Under 5GB is a GGUF model optimized to run entirely in VRAM on a 6GB GPU, achieving 65 tokens per second. It provides a lightweight local coding model option for constrained hardware.
Why it’s interesting
It runs fully in VRAM on a modest 6GB GPU with high throughput under 5GB size.
Target user
Developers and users with low-spec GPUs running local coding models
Source metrics: Points 3 · Comments 0
HN discussion · Project
Source: #HackerNews / Show HN