**thedotmack/claude-mem** (⭐110k) is a persistent memory-compression system for agents; today's news is less a feature than reach — it now supports Grok Bot, T3 Code, Antigravity CLI, OMP, Pi, and DeepSeek Harness, covering nearly every coding agent you can name. **max-sixty/worktrunk** (⭐9.9k) is a Rust CLI that collapses git worktree management into three commands, fixing the friction of typing a branch name three times just to spin up a new workspace for a parallel agent. **db-lyon/ue-mcp** (⭐466) exposes the entire Unreal Editor as MCP tools — of its 2,114+ actions, 830 call Epic's own native AI Toolset directly. **FlashML-org/FreeToken** (⭐16k) is an edge-native MoE inference engine aiming to run 290B+ frontier open-weight models at interactive speed on an ordinary gaming PC. No major framework release today.
A Mac has no NVIDIA GPU, so it can't really run vLLM — but Colab's free T4 (compute capability 7.5, exactly vLLM's minimum) can. Serve the 0.6B Qwen3 as an OpenAI-compatible API with vllm serve, fire 8 requests at once, read /metrics, and you can watch continuous batching happen.
11-868 spends two lectures on one question: how does an inference server handle many requests at once without wasting KV cache on the GPU? Lecture 22 (Lei Li) starts from SGLang's scheduling loop: ORCA's continuous batching, RadixAttention's radix tree for KV, sorting and routing by prefix hit rate, and hiding CPU scheduling behind GPU compute. Lecture 24 is given by vLLM author Woosuk Kwon: PagedAttention cuts KV cache into fixed-size blocks and virtualizes them with a block table, taking the batch on one A100 from 8 to 40. The second half covers how vLLM cuts CPU overhead, uses piecewise CUDA graphs, splits models across GPUs, and manages memory for hybrid architectures.
11-868 closes with five serving decks: Hao Zhang on DistServe, Vikram Mailthody on NVIDIA Dynamo, Junchen Jiang on LMCache, Mingxing Zhang on Mooncake and KTransformers, and Lei Li's map of serving frameworks. They share one question: once serving grows from one machine to a data center, where do the compute and the KV cache go? The argument runs in three steps. Measure goodput under latency SLOs instead of raw throughput. Put prefill and decode on separate GPUs. Let the KV cache spill from GPU memory into CPU memory, SSDs, and remote storage.
Lecture 13 sorts the ways to speed up LLM inference into three paths. Quantization: SmoothQuant moves the difficulty of activation outliers onto the weights to make W8A8 work, AWQ uses activation magnitudes to find the roughly 1% of weights that matter and protects them by scaling for W4A16, and QServe combines both into W4A8KV4. Sparsity: Wanda prunes weights by |W|·‖X‖, DejaVu and MoE use only part of the parameters per token, and SpAtten and H2O drop unimportant tokens. Serving: TTFT/TPOT metrics, PagedAttention, FlashAttention, speculative decoding, and continuous batching. On the slides, INT3 OPT-6.7B has a perplexity of 43.16 with RTN; scaling the salient channels by 2 brings it to 14.07.
learn-inference.com is an unofficial interactive companion to Philip Kiely's Inference Engineering (256 pages, Baseten Books, free PDF). It follows the book's 8 chapters and 42 sections with rewritten explanations, turns intuition-heavy ideas like TTFT, P99, speculative decoding, and prefix-cache routing into slider-driven simulators, and ships a keyless JSON API and MCP server.
Xinference wraps vLLM, SGLang, llama.cpp, Transformers, and MLX under a single management layer, using a Web UI and OpenAI-compatible API to manage LLMs, embedding, rerank, speech, and image models — suited for self-hosted deployments that need multiple model types to coexist. But the management layer's parsing logic also creates a larger attack surface than pure serving engines (CVE-2026-61539 is a case study).
Fireworks AI puts open-weight model evaluation, dedicated GPU deployments, and LoRA customization behind one API surface. Serverless fits low-volume starts, On-demand fits sustained traffic and custom models, while reserved capacity adds enterprise capacity guarantees.
Triton Inference Server serves TensorRT, ONNX, PyTorch, and other models through consistent HTTP and gRPC APIs. Its defining tools are the model repository, dynamic batching, instance groups, and ensembles—not LLM-specific KV-cache scheduling.
CS25 is Stanford's 1-unit seminar where attendance is the only homework and anyone can audit. Of the nine talks in the Spring 2026 season, the three worth your time are Albert Gu on the inductive biases of SSMs vs Transformers, Charles Frye on serving inference across thousands of GPUs, and Victoria Lin on what native multimodality still hasn't solved.
vLLM uses PagedAttention to eliminate KV cache memory waste, combining continuous batching and prefix caching to become the most widely adopted open-source LLM inference engine today.