Aug 24 to Aug 30, 2026

peonist-ai

@peonist-ai · Aug 26

Container-based server tuned for the gfx1151 Strix Halo APU, aimed at squeezing maximum tokens per second from the 125B model on a mini-PC.

slee-persis

@slee-persis · Aug 26

Ensemble harness claiming a group of small Qwen3.8 models reach frontier-level coding scores at a fraction of the cost. Results are self-reported and worth verifying.

Derpyu520

@Derpyu520 · Aug 25

Bridges QQ (via OneBot v11) to DSH agents with social simulation, slang learning and safe MCP tools. The author built it so an agent can behave like a regular member of a chat group.

Alok

@analogalok · Aug 26

Posts a video of the 125B MoE running at about 21 tok/s decode with a quarter-million-token window on one gaming card, no quantised cache.

Ten-minute walkthrough installing DeepSeek Harness as a free, self-owned Claude Code alternative; the biggest tutorial of the launch.

OnlyTerp

@OnlyTerp · Aug 27

Open-source project opengrok: Run any model in Grok Bot — one-command setup, model picker UI, evidence-based provider wire maps, and an update-proof d. Extends or complements Grok Bot for builders.

carloslfu

@carloslfu · Aug 28

Streams expert weights from SSD so a Mac that cannot hold the model in memory still runs it at roughly 12 tok/s. A neat trick for memory-poor Apple machines.

blazux

@blazux · Aug 27

Early write-up of how to serve the model on a single DGX Spark, published within a day of the weights landing.

MiaAI-Lab

@MiaAI-Lab · Aug 26

Dual-node vLLM deployment of the NVFP4 build across a pair of DGX Sparks, for higher context and throughput.

Unsloth AI

@UnslothAI · Aug 26

Launch-day announcement of GGUF quants and a guide for CPU and unified-memory setups.

acryldev

@acryldev · Aug 25

A context-relay workspace that keeps a single canonical context while handing work between different agents including DeepSeek Harness. Built to stop people copy-pasting state between tools.

Soodok

@Soodok · Aug 26

Packages the DSH agent as an Android app with an extension center and no root requirement. Built by a 15-year-old developer so the agent works on a phone with nothing else installed.

tonyd2wild

@tonyd2wild · Aug 26

VLLM patch that keeps the huge n-gram table on disk and reaches about 44 tok/s on one Spark.

0xBakeer

@0xBakeer · Aug 26

Shows how paging the embedding table to NVMe lets the 180B-class checkpoint run on a single DGX Spark.

DominikBucko

@DominikBucko · Aug 30

Dual-3090 build with 128GB RAM reporting about 4,200 tok/s prefill and 111 tok/s decode on long prompts.

runta-dev

@runta-dev · Aug 26

Open-source take on Grok Bot where persistent AI teammates each get a cloud computer, powered by Runta.

majiayu000

@majiayu000 · Aug 30

Open-source project awesome-grok-bot: 3387 live x.ai/bot shares for Grok Bot — curated catalog with shelf tiers, searchable Pages, bilingual README. Not a pro. Extends or complements Grok Bot for builders.

opencoredev

@opencoredev · Aug 27

Open-source take on named AI teammates with desktop, web and mobile clients and per-bot tools, instructions and memory.

Dyluhn

@Dyluhn · Aug 30

Hand-tuned ROCm inference path for AMD RDNA4 cards running the 125B model.

opdsh

@opdsh · Aug 28

Lets the agent drive the Unity Editor with the unity CLI for scene and asset tasks. Meant for game developers who want agent help inside the editor.

Brian Casel

@CasJam · Aug 29

Brian Casel reorganised his portfolio so each product has a head bot, growth bot and other roles reporting to a chief bot.

sxuff

@sxuff · Aug 26

Tested llama.cpp GGUF route for a single Spark or Asus GX10 box.