RTX 5090 Benchmarks Qwen3.8 27B: Stable Speed at Long Context
_-_David · reddit · 2026-08-18
Real-world performance data for Qwen3.8 27B on an RTX 5090 based on 7 hours of continuous testing (1,274 generations). Median decode speeds remained surprisingly stable as context grew:
| Prompt Size | Decode Speed |
| :--- | :--- |
| <50k | 169.7 tok/s |
| 50–100k | 167.6 tok/s |
| 100–150k | 156.9 tok/s |
| 150–200k | 149.0 tok/s |
| 200k+ | 143.7 tok/s |
Peak Metrics:
- Max Decode (Single Request): 222.0 tok/s
- Max Decode (Sustained): 211.2 tok/s
- Config: KV cache set to q8, max context length, stock clocks.
More from Infra
- Large data centers employ only ~50-150 permanent on-site staff, Microsoft says — brezina · 2026-10-16
- Prediction: Opus 5.5-level models on laptops by 2027, backed by Qwen 27B data — cephaloform · 2026-10-16
- Oracle down 12% in a week, nearly 50% off June high as OpenAI exposure spooks market — alvelda · 2026-10-16
- US data center moratoriums top 400 — an AI bull explains why they're actually good for AI — kevinsxu · 2026-10-16
- Microsoft Research unveils Magnetar, the fastest zkVM at 10-39x speedups and post-quantum security — jedisct1 · 2026-10-16
- Bonsai 2 2B hits 120 tok/s on iPhone 18 Pro, outpaces Qwen 3.5 2B at half the memory — cephaloform · 2026-10-16