Muse-Glimmer-30B-ctx48k-kvq8 on RTX 4090 (24GB)
MeasuredOn 2026-10-22, Muse-Glimmer-30B-ctx48k-kvq8 UD-Q4_K_XL ran at 52.7 tok/s (chat-short) on an RTX 4090 (24GB) with llama.cpp. Run r_7srd_wfvix-.
Workload results
| Workload | Backend | Model | decode tok/s | prefill tok/s | TTFT | p50 | p95 |
|---|---|---|---|---|---|---|---|
| chat-short | llama.cpp@0.7.0-dev (build 11803, commit a37cc4610) | Muse-Glimmer-30B-ctx48k-kvq8UD-Q4_K_XL | 52.71tok/s | 614.6tok/s | 251ms | 18.9ms | 19.1ms |
| chat-long | llama.cpp@0.7.0-dev (build 11803, commit a37cc4610) | Muse-Glimmer-30B-ctx48k-kvq8UD-Q4_K_XL | 51.08tok/s | 2,743.7tok/s | 1,144ms | 19.5ms | 19.8ms |
| concurrent-decode | llama.cpp@0.7.0-dev (build 11803, commit a37cc4610) | Muse-Glimmer-30B-ctx48k-kvq8UD-Q4_K_XL | 52.28tok/s | — | — | 19.1ms | 19.4ms |
| agent-trace | llama.cpp@0.7.0-dev (build 11803, commit a37cc4610) | Muse-Glimmer-30B-ctx48k-kvq8UD-Q4_K_XL | 50.95tok/s | 5,986.9tok/s | 358ms | 19.4ms | 19.6ms |
Reproduce on your machine
Start with this model and workload selection. This command does not capture the original artifact, backend version or server settings:
Match quantization, context capacity, caching and runtime settings for a useful comparison. Duration varies by workload and hardware; the published client may differ from this run's version. How it's measured.
Share this run
Copy the short link, a Reddit-ready summary, or a README badge. Every number is the measured result on this page, and each badge links back to it.
https://llm-speed.com/r/r_7srd_wfvix-Related benchmarks
- More Muse-Glimmer-30B-ctx48k-kvq8 benchmarks — every backend and rig that has run this model.
- More RTX 4090 (24GB) LLM benchmarks — every model measured on this hardware.
Provenance
- Run ID
- r_7srd_wfvix-
- Fingerprint hash
- Public key
- 1JuyY+jqPHSZWQaDryGKIonLn4Nj3I5JtQVl0e518sI=
- Received
- 2026-10-22 10:40:16