32 GB · 1,792 GB/s
What can the NVIDIA RTX 5090 run?
The fastest single card you can buy, at a very high 2026 price.
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 greatQ4_K_M · needs 19.9 GB at 8K · 55–90tok/sFast
- Gemma 4 31BRuns greatQ4_K_M · needs 21.4 GB at 8K · 55–85tok/sFast
- Muse Glimmer 30BRuns greatQ4_K_M · needs 19.8 GB at 8K · 60–95tok/sFast
Coding21 of 26 run
- Qwen3.8 27BRuns greatQ4_K_M · needs 21.5 GB at 32K · 55–85tok/sFast
- Muse Glimmer 30BRuns greatQ4_K_M · needs 20.2 GB at 32K · 55–90tok/sFast
- Qwen3.6 35B-A3BRuns wellQ4_K_M · needs 24.9 GB at 32K · 130–260tok/sFast
Long documents15 of 20 run
- Qwen3.8 27BRuns greatQ4_K_M · needs 21.5 GB at 32K · 55–85tok/sFast
- Qwen3.6 35B-A3BRuns wellQ4_K_M · needs 24.9 GB at 32K · 130–260tok/sFast
- Gemma 4 12BRuns greatQ8_0 · needs 14.7 GB at 32K · 75–120tok/sFast
Agents & tools21 of 26 run
- Qwen3.8 27BRuns greatQ4_K_M · needs 21.5 GB at 32K · 55–85tok/sFast
- Qwen3.6 35B-A3BRuns wellQ4_K_M · needs 24.9 GB at 32K · 130–260tok/sFast
- Ornith 1.5 35B-A3BRuns greatQ4_K_M · needs 23.4 GB at 32K · 130–260tok/sFast
Images in9 of 13 run
- Qwen3.8 27BRuns greatQ4_K_M · needs 19.9 GB at 8K · 55–90tok/sFast
- Qwen3.6 35B-A3BRuns wellQ4_K_M · needs 24.4 GB at 8K · 150–290tok/sFast
- Muse Glimmer 30BRuns greatQ4_K_M · needs 19.8 GB at 8K · 60–95tok/sFast
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
Qwen3.6 35B-A3B (Q4_K_M, 22.3 GB file) at about 150–290tok/sFast.
Should you buy it?
Buy it if
You want one card that runs every 30B-class model fast, with room for long conversations.
Skip it if
You can live with used hardware: two RTX 3090s give 48 GB for far less money.
Watch out for
- 575 W: needs a 1,000 W+ power supply and good airflow.
- Prices are about three and a half times the $1,999 launch price; check before you buy.
Measured on this hardware
Real runs from public sources, not estimates. All measurements.