OpenJet claims Qwen3.8-27B coding runs 1.9x faster than llama.cpp on one RTX 3090
JeremyCMorgan · x · 2026-10-23
L-Forster's open-source OpenJet engine reports 140 tokens/s for coding with Qwen3.8-27B on a single RTX 3090, versus 73 tokens/s with llama.cpp — a 1.9x speedup on 24GB+ RTX 30/40/50-series cards. The offline coding agent sets up the best model your GPU can run with one command. Engine code and benchmarks are public (AGPL-3.0), and the author is asking the community to reproduce the results.
More from coding & agent
- We tested 12 LLM classifiers — the new classifier hype is LARP — KlausCodes · 2026-10-23
- Dev builds ESP32 desk display showing Claude Code and Codex agent status via local Go helper — tonylook · 2026-10-23
- Let your agent read its own traces: AgentsView + history tool debugging workflow — dbreunig · 2026-10-23
- YC-backed Mireye launches land search API + MCP with 160M parcels for AI agents — ycombinator · 2026-10-23
- Testing 12 decision systems: the new LARP hype around classifiers — KlausCodes · 2026-10-23
- Monid launches 'Product Hunt for agents,' ranking tools by real agent calls — claudeai · 2026-10-23