Atlas Inference is a dependency free lightweight inference engine meant to run LLMs on YOUR compute seamlessly. Supports GB10 and Strix Halo at the moment, hoping to expand to more!
☆34Oct 18, 2026Updated this week
Alternatives and similar repositories for atlas
Users that are interested in atlas are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- English guide + one-command patcher for the vLLM spin-wait bug that overheats NVIDIA DGX Spark / GB10. Credit: nacyot. CPU cores spin at …☆24Aug 17, 2026Updated 2 months ago
- vLLM + Qwen3.6-27B (BF16) OpenAI-compatible inference server on AMD Strix Halo (Ryzen AI Max+ 395, gfx1151). Vision input, 256K context, …☆18Apr 26, 2026Updated 5 months ago
- A tiered-memory system design for workloads that don't fit in RAM: measure the working set, pin the hot tier, stream the cold tier from f…☆28Updated this week
- Serve the official Ling-3.0-flash INT4 on one NVIDIA DGX Spark, 38.7 tok/s, faster than the community GGUF. Recipe, watchdog, benchmarks.☆26Aug 29, 2026Updated last month
- CUDA Blackwell performance fork of antirez/ds4 (DeepSeek-V4-Flash): ~4x prefill, ~1.5x decode, continuous batching, weight server, DSpark☆77Sep 4, 2026Updated last month
- Bare Metal GPUs on DigitalOcean Gradient AI • AdPurpose-built for serious AI teams training foundational models, running large-scale inference, and pushing the boundaries of what's possible.
- An GB10-optimized inference engine☆108Updated this week
- Optimized inference for the DGX Spark and spark clusters☆123Updated this week
- Minimal read-only OS image for GB10 (DGX Spark / ASUS GX10) boxes☆89Oct 4, 2026Updated 2 weeks ago
- vLLM inference solutions for NVIDIA GB10 (DGX Spark, sm121) — KIMI-K3 B12X_MLA + DSpark, GLM-5.2☆67Sep 30, 2026Updated 2 weeks ago
- Container toolbox that runs llama.cpp with AMD ROCm and NVIDIA CUDA in one build, for inference on two GPUs☆20Updated this week
- Proxima lets existing GPUs serve 4x more concurrent requests☆26Aug 19, 2026Updated last month
- DeepSeek-V4-Flash-DSpark optimized vLLM 0.26.0 SM121 dual-GB10 evidence (NVFP4 KV, B12X, DSpark K6)☆31Aug 3, 2026Updated 2 months ago
- ☆27Sep 11, 2026Updated last month
- Event-driven long-horizon memory for real-time wearable AI and LLM agents.☆30Aug 18, 2026Updated 2 months ago
- Virtual machines for every use case on DigitalOcean • AdGet dependable uptime with 99.99% SLA, simple security tools, and predictable monthly pricing with DigitalOcean's virtual machines, called Droplets.
- Switchless RoCE collectives, NCCL profiles and fabric tooling for DGX Spark inference recipes☆37Updated this week
- Autonomous self-improving 4x DGX Spark (GB10) MoA stack + LoRA loop (DSV4F router, Qwen3.6/Omni/TwoTower/Gemma). Hermes MoA routing, ~90%…☆24Aug 17, 2026Updated 2 months ago
- GLM-5.3-Flash NVFP4 on 4x NVIDIA DGX Spark: vLLM TP4, DFlash2 adaptive draft, measured FP8 vs NVFP4 vs EXL3 vs SGLang☆37Sep 29, 2026Updated 2 weeks ago
- This is a mirror of the Strix Halo HomeLab wiki, to browse the wiki click on the link below☆73Aug 31, 2026Updated last month
- keys-DGX-Sparkticus Utimate Power Pack Unleashed (Dual DGX-Sparks Needed)☆43Aug 23, 2026Updated last month
- P2P support for nvidia consumer cards☆23Sep 5, 2026Updated last month
- Strata fork: Qwen3.8-Flash-Next 125B MoE in NVFP4 on one RTX 20-50 card (12 GB+) and 64 GB of RAM or more. Setup installs our GPTQ quants…☆165Updated this week
- ENGRAFT: write facts into an LLM's n-gram (Engram) memory table as a removable overlay: 100 facts in a 9 MB file, no weights touched, ver…☆56Sep 30, 2026Updated 2 weeks ago
- DGX Spark (GB10) random hard power-off: how to diagnose the log-less crash signature and stop it with a GPU clock cap. Field-tested fix +…☆25Aug 16, 2026Updated 2 months ago
- Deploy on Railway without the complexity - Free Credits Offer • AdConnect your repo and Railway handles the rest with instant previews. Quickly provision container image services, databases, and storage volumes.
- ☆90Oct 1, 2026Updated 2 weeks ago
- Remove image backgrounds locally, in the browser or from the terminal. Nothing is uploaded.☆26Sep 3, 2026Updated last month
- RP2040 based iPhone (lightning) UART cable, that actually works (hardware)☆37Aug 22, 2026Updated last month
- Community-tested recipes, diagnostics, and benchmarks for CMP 170HX AI workloads☆84Oct 10, 2026Updated last week
- ☆29Jun 10, 2026Updated 4 months ago
- 解锁自有 NVIDIA CMP 170HX / 90HX 的算力限制。Linux x86-64，需要 root。☆57Aug 6, 2026Updated 2 months ago
- Community registry of LLM serving-path traps that produce confidently wrong measurements: templates, tool parsers, reasoning fields, quan…☆135Updated this week
- Route every page to the right parser☆27Jul 30, 2026Updated 2 months ago
- qwen38-flash-next recipe for tp=2 GB10☆127Oct 6, 2026Updated last week
- GPUs on demand by Runpod - Special Offer Available • AdRun AI, ML, and HPC workloads on powerful cloud GPUs—without limits or wasted spend. Deploy GPUs in under a minute and pay by the second.
- Kyojin: the Yamz inference engine for AMD Strix Halo (ROCm, gfx1151), built on ExLlamaV3. Runs 300B-class MoE models on one 128 GB mini P…☆189Updated this week
- Galahad by Corbenic☆48Oct 4, 2026Updated 2 weeks ago
- A 4B decision model: calibrated probabilities for yes/no, choice and score questions over long inputs.☆38Updated this week
- Qwen3.8-27b and Qwen3.6-35B-A3B local inference engine for Vulkan and Metal☆71Updated this week
- Make your agents more efficient with a precomputed map of your Git repository: ranked code search plus definitions, callers, types, ro…☆57Updated this week
- Performance-tuned llama.cpp for AMD Strix Halo (gfx1151): FA + MoE-prefill fixes with a bundled current Mesa driver. Vulkan and HIP; port…☆217Oct 9, 2026Updated last week
- Run DeepSeek-V4-Flash at full 1M context AND two MiniMax H3 video instances on the same two DGX Sparks. Benchmarked C1-C6.☆35Aug 7, 2026Updated 2 months ago