# Inference: what I've logged

> Inference comes up in 2 posts of my AI log, from October 2026 to October 2026, and each item below links to the primary source. The sources come from MemTensor and DeepSeek, filed under agents and coding, open weights and frontier.

## Items

1. **MemT-1.0: Towards Memory-Native Evolution** (MemTensor, 24 Oct 2026): Paper on building memory into the model with Gated DeltaNet-2, reporting about 1.86x inference speedup and 66x less persistent storage than a full-context KV cache at 128K tokens. [Primary source](https://arxiv.org/abs/2610.26825) · [Post](https://heyaliasif.com/ai/2026-10-25-gvs5h-cerebras-trillion-memt#item-3)
2. **DeepInfer** (DeepSeek, 19 Oct 2026): The inference framework DeepSeek says powers its production services, released under Apache 2.0 for offline inference with DeepSeek-V4.1-Flash. [Primary source](https://github.com/deepseek-ai/DeepInfer) · [Post](https://heyaliasif.com/ai/2026-10-20-qwen38-flash-next-deepinfer-pytorch-mps#item-2)

## Posts

- 2026-10-25 [GVS5H self-orchestration, Cerebras at trillion scale, MemT-1.0](https://heyaliasif.com/ai/2026-10-25-gvs5h-cerebras-trillion-memt)
- 2026-10-20 [Qwen3.8-Flash-Next, DeepSeek's DeepInfer, PyTorch 2.15 on MPS](https://heyaliasif.com/ai/2026-10-20-qwen38-flash-next-deepinfer-pytorch-mps)

Canonical: https://heyaliasif.com/ai/topics/inference