LinearSolveBench debuts: testing if AI can write fast C solvers for sparse linear systems
hgarud · reddit · 2026-09-22
Redditor hgarud released LinearSolveBench, a new open-source benchmark on GitHub.
- Measures a model or harness's ability to write fast, accurate, and general numerical solvers
- Tasks: implementing C solvers for large sparse linear systems
- Goal: spur algorithmic advances in numerical methods, not just library calls
A benchmark at the intersection of numerical algorithms and code generation, relevant for anyone probing the limits of agentic coding.
More from coding & agent
- EvolveTrade Paper: Evolving Agent System Prompts Beats Fixed-Prompt Trading Baselines — dair_ai · 2026-09-23
- DigitalOcean Launches Managed Agents in Public Preview, Hosting Claude Code, Codex and LangGraph — SimplyAnnisa · 2026-09-23
- StepFun's Step Code ships with one-line install and built-in StepPage publishing — StepFun_ai · 2026-09-23
- StepFun open-sources Step Code: 80.9% on Terminal-Bench 2.1 with fewer tokens — StepFun_ai · 2026-09-23
- Adaptive UIs mean mobile devs must verify functions, not outputs — rseroter · 2026-09-23
- X exec: bot detection will be an urgent business as agent swarms flood the web — nikitabier · 2026-09-23