Iluvatar CoreX Tiangai 300 tops Nvidia H200 in select AI inference tests, with long-context and MoE gains

modelsDigiTimesNov 2, 2026

Chinese AI chipmaker Iluvatar CoreX has released benchmarks showing its next-generation Tiangai 300 general-purpose GPU outperforming Nvidia's H200 in selected large language model (LLM) inference, long-context processing and multimodal generation workloads.

Impact: LowConfidence: LowNovelty: Incremental
modelsdigitimesmultimodal

See the whole story

Read the original on DigiTimes

Sens.ai aggregates and assesses published reporting. The assessment above is machine generated; the original source is authoritative.