Artificial Analysis launches benchmarks for small models on mobile phones
ArtificialAnlys · x · 2026-08-25
Artificial Analysis, in partnership with Liquid AI, has launched intelligence and inference benchmarks for small models on mobile phones. The intelligence benchmark uses five evaluations tailored for function calling, knowledge, and reasoning tasks. The inference benchmark measures speed, latency, and memory on real devices. The article highlights that benchmarking small models is difficult due to the clustering of scores on traditional frontier benchmarks and the immaturity of mobile runtimes. Tests use 4-bit or smaller quantized builds.
Related event: Artificial Analysis and Liquid AI Launch On-Device Small Model Benchmarks(8 posts)→
More from Infra
- Mistral partners with Saudi HUMAIN to build localized AI models and infrastructure — MistralAI · 2026-08-25
- Lovable hits 250 repos/sec peak using Code.Storage for AI coding infra — dhruv2038 · 2026-08-25
- SpaceX's space data center concept simplifies to chips, power, and cooling — XFreeze · 2026-08-25
- NVIDIA splits agentic AI workloads across Rubin GPUs, Groq LPX and Vera CPUs — rohanpaul_ai · 2026-08-25
- AI Data Centers Face Local Resistance: Jobs Needed as Energy Costs Rise — dbasch · 2026-08-25
- Nvidia spending $6B to build US alternative to Chinese AI — abhishekcode42 · 2026-08-25