PC without a graphics card, 8 GB RAM. Updated 2 November 2026.
PC without a graphics card, 8 GB RAM can run 2 of the 31 models tracked here at a comfortable speed, and another 3 slowly. No chat model fits comfortably, so this machine is best kept for speech tools and tiny assistants.
Without a graphics card everything runs on the processor. System RAM (8 GB here) holds the model and moves data at roughly 70 GB/s, so even small models produce only a few words a second. It works for small assistants and speech tools, and is not realistic for image or video generation.
Best chat pick: Gemma 4 E2B at about 14 words a second. Best coding pick: Qwen2.5 Coder 7B at about 6.
Image models need more memory than this machine has.
Video models need more memory than this machine has.
No text model runs at a useful speed.
Nothing fits comfortably, so there is no headroom to speak of.
The biggest single jump is PC with RTX 3090 (24 GB) and 8 GB RAM, which would unlock 21 more models, including Gemma 4 E2B, Qwen 3.5 4B, Gemma 4 E4B.
For chat, Gemma 4 E2B is the best fit, at about 14 words a second. For coding, Qwen2.5 Coder 7B.
Gemma 4 12B does not fit; gpt-oss 20B does not fit; Qwen 3.5 35B A3B does not fit.
No. DeepSeek R1 8B needs about 6 GB and this machine tops out near 6 GB.
| Model | Verdict | Speed | Needs |
|---|---|---|---|
| Whisper large-v3 turbo | Runs well | about 10 min | 4 GB |
| Kokoro TTS | Runs well | about 2 min | 2 GB |
| Qwen2.5 Coder 7B | Slow | 6 words/s | 5.4 GB |
| Gemma 4 E2B | Slow | 14 words/s | 5.3 GB |
| Qwen 3.5 4B | Slow | 9 words/s | 4 GB |
As an Amazon Associate I earn from qualifying purchases.
Unlocks 12 more: Gemma 4 E2B, Qwen 3.5 4B, Gemma 4 E4B, Qwen 3.5 9B and more.
Unlocks 14 more: Gemma 4 E2B, Qwen 3.5 4B, Gemma 4 E4B, Qwen 3.5 9B and more.
Unlocks 21 more: Gemma 4 E2B, Qwen 3.5 4B, Gemma 4 E4B, Qwen 3.5 9B and more.