DeepSeek v4.1 Flash Added to LiveBench Leaderboard
LiveBench has added DeepSeek v4.1 Flash to its LLM evaluation leaderboard. The benchmark results indicate that the model achieves performance on par with leading models at less than one-tenth of the evaluation cost. This update demonstrates that low-cost AI models are increasingly capable of matching top-tier proprietary models in complex benchmark tasks. It enables developers and enterprises to access strong AI reasoning capabilities at drastically lower operational expenses. DeepSeek v4.1 Flash achieves comparable scores to high-ranking models while maintaining a fraction of the cost per successful task on LiveBench evaluations. LiveBench measures these scores across objective tasks to ensure fair performance and cost comparisons.
## BACKGROUND
LiveBench is an LLM benchmark sponsored by Abacus.AI designed to prevent test set contamination through frequently updated, objectively evaluated problem sets. DeepSeek is an AI organization recognized for producing highly efficient frontier and open-source models that deliver competitive benchmarks at lower computing costs.