48 GB · 936 GB/s
What can Two RTX 3090s (used) run?
48 GB of graphics memory for less than one new 32 GB card. Loud and power-hungry.
Best models for each task
Our top three for this machine. Chat and image tasks assume an 8K-token conversation; coding, documents and agents assume 32K.
Chat & writing20 of 25 run
- Qwen3.8 27BRuns greatQ8_0 · needs 31.6 GB at 8K · 19–30tok/sSmooth
- Gemma 4 31BRuns wellQ8_0 · needs 35.7 GB at 8K · 17–25tok/sSmooth
- Muse Glimmer 30BRuns greatQ8_0 · needs 32.1 GB at 8K · 19–30tok/sSmooth
Coding21 of 26 run
- Qwen3.8 27BRuns greatQ8_0 · needs 33.2 GB at 32K · 18–30tok/sSmooth
- Muse Glimmer 30BRuns greatQ8_0 · needs 32.5 GB at 32K · 19–30tok/sSmooth
- Qwen3.6 35B-A3BRuns wellQ8_0 · needs 40.4 GB at 32K · 70–140tok/sFast
Long documents15 of 20 run
- Qwen3.8 27BRuns greatQ8_0 · needs 33.2 GB at 32K · 18–30tok/sSmooth
- Qwen3.6 35B-A3BRuns wellQ8_0 · needs 40.4 GB at 32K · 70–140tok/sFast
- Gemma 4 26B-A4BRuns greatQ8_0 · needs 29.9 GB at 32K · 60–110tok/sFast
Agents & tools21 of 26 run
- Qwen3.8 27BRuns greatQ8_0 · needs 33.2 GB at 32K · 18–30tok/sSmooth
- Qwen3.6 35B-A3BRuns wellQ8_0 · needs 40.4 GB at 32K · 70–140tok/sFast
- Ornith 1.5 35B-A3BRuns wellQ8_0 · needs 39.5 GB at 32K · 70–140tok/sFast
Images in9 of 13 run
- Qwen3.8 27BRuns greatQ8_0 · needs 31.6 GB at 8K · 19–30tok/sSmooth
- Qwen3.6 35B-A3BRuns wellQ8_0 · needs 39.9 GB at 8K · 75–150tok/sFast
- Muse Glimmer 30BRuns greatQ8_0 · needs 32.1 GB at 8K · 19–30tok/sSmooth
Make images6 of 6 run
- Stable Diffusion XL Base 1.0Runs greatFP16 · needs 7.9 GB
- FLUX.2 [klein] 4BRuns greatBF16 · needs 14.0 GB
- Z-Image-TurboRuns greatBF16 · needs 17.0 GB
Speak text7 of 7 run
- Qwen3-TTS 1.7BRuns greatBF16 · needs 5.5 GB
- Kokoro-82MRuns greatFP32 · needs 1.4 GB
- Chatterbox Multilingual V3Runs greatFull · needs 4.2 GB
Transcribe7 of 7 run
- Voxtral Mini 4B RealtimeRuns greatBF16 · needs 17.0 GB
- Whisper large-v3-turbo (via faster-whisper)Runs greatFP16 · needs 2.6 GB
- Nemotron 3.5 ASR Streaming 0.6BRuns greatFP32 · needs 3.5 GB
Biggest model it runs well
Kimi Linear 48B-A3B (Q4_K_M, 30.1 GB file) at about 110–220tok/sFast.
Should you buy it?
Buy it if
You want 70B models or long context on a budget, and have a big case and power supply.
Skip it if
You want silence, a small box, or simple setup.
Watch out for
- Needs a 1,200 W+ power supply.
- Speed: the model is split across both cards, so it runs at about one card’s speed.
Measured on this hardware
Real runs from public sources, not estimates. All measurements.