linear-attention
Agnes 3.0 Flash's open weights are a 33B preview, not the model behind its leaderboard score News
Agnes AI released open weights for a 33-billion-parameter multimodal Agnes 3.0 Flash under Apache 2.0 on 11 September 2026, but its model card says these are an earlier preview checkpoint and that the leaderboard score circulating with them belongs to a different production model.
Ling-3.0 ships a hybrid-linear mixture of experts under a plain MIT license News
inclusionAI released the Ling-3.0 family on Hugging Face with a permissive MIT license, mixing linear and full attention in a fixed three-to-one ratio, and the smallest variant activates only 1.3 billion of its 7.9 billion parameters per token.
Ling-3.0 Tool
inclusionAI's hybrid-linear mixture-of-experts family under a plain MIT license, mixing three linear-attention blocks per full-attention block across 128 routed experts. The tiny variant holds 7.9B parameters and activates 1.3B per token; native BF16, FP8 and INT4 support is declared on the card.
FlashKDA Tool
Moonshot AI's MIT-licensed kernel implementation of Kimi Delta Attention, the linear-attention mechanism underneath Kimi K3, published ahead of the model weights themselves. Useful today for anyone building or serving bounded-state attention rather than a growing key-value cache.
Agnes-3.0-Flash (Preview) Tool
Agnes AI's 33-billion-parameter text, image and video model, released 11 September under Apache 2.0, with a 262,144-token context and a hybrid recurrent-plus-attention design. A 66.2 GB bf16 download; the model card asks for one NVIDIA H200 (141 GB) or H100 (80 GB) plus 128 GB of host memory, and a community Q4_K_M GGUF is 19.75 GB. It is a Preview checkpoint: the Artificial Analysis score belongs to a different API model.