Measurement wanted · request DR-016 · opened 2026-11-06
Ripple-65B-Preview on RTX 3090: related runs (other quant or GPU)
We don't have a Ripple-65B-Preview speed run on RTX 3090 yet, so there is no tokens-per-second number here. People asked for it, and this page is where the result will appear. Runs of the same model on another quant or GPU are listed below; see them all on the Ripple-65B-Preview model page. Anyone with that hardware can run it with the free llm-speed CLI, and we'll post it as a community run.
What's being requested
- Model
- Ripple-65B-Preview · ricursiveAI/Ripple-65B-Preview
- Hardware
- RTX 3090
- Also requested
- RTX 4090, RTX 5090
Runs we already have
Related runs (other quant or GPU)
- r__2bhvnz5xs9 · RTX 5090 (32GB) · 56.3 tok/s decode (chat-short) · Community-submitted
- r_1_clgx0ov_8 · RTX 5090 (32GB) · 3.5 tok/s decode (chat-short) · Community-submitted
- r__jth1z3c93v · RTX 5090 (32GB) · 42.1 tok/s decode (long-context-decay) · Community-submitted
- r_q565ikt_b7z · RTX 5090 (32GB) · 44.8 tok/s decode (chat-short) · Community-submitted
- r_t7dhb1ohnpz · RTX 5090 (32GB) · 45.0 tok/s decode (chat-short) · Community-submitted
- r_vhoo1k08gx- · RTX 4090 (host-reported 48GB) · 35.6 tok/s decode (chat-short) · Community submission · provenance unverified
- r_kjxer-wvywz · RTX 4090 (host-reported 48GB) · 38.5 tok/s decode (chat-short) · Community submission · provenance unverified
- r_7ktcs0gslo4 · RTX 4090 (host-reported 48GB) · 38.6 tok/s decode (chat-short) · Community submission · provenance unverified
- r_o83dygfgzlq · RTX 4090 (host-reported 48GB) · 38.6 tok/s decode (chat-short) · Community submission · provenance unverified
Run it yourself
A stock run needs llama.cpp built with Ripple support (merged in PR #31692, not yet in a tagged release). Pull request.
Start llama-server with the model (flags below are a starting point; adjust offload to your machine), then point llm-speed at it. Add --llama-args='...' with the flags you used so the run records them.
pipx install llm-speed # or: pipx upgrade llm-speed
# llama.cpp built with Ripple support (merged in PR #31692, not yet in a tagged release)
llama-server -m /path/to/MODEL.gguf --port 8080 -ngl 999
# then, in a second terminal:
LLAMA_CPP_SERVER_URL=http://127.0.0.1:8080 \
llm-speed bench --backend llama.cpp \
--model /path/to/MODEL.gguf \
--workload chat-shortThe upload is signed and public. How contributing works · what the CLI sends.
Get an alert
Email alerts for this page are switching on soon. Until then, follow @llmspeed on X, where new runs are posted.