janas
Janas-LLM: large mixture-of-experts language models on ordinary computers, in C. Experts streamed from NVMe, with or without a GPU.
Details
- External ID
- 1378187980
- Source
- GITHUB
- Company
- —
- Product
- janas
- Website domain
- github.com
- Launched
- Sept. 20, 2026
- Cohort
- —
- Upvotes
- 21
- Upvotes percentile
- 0.6211247758134768
- Tags
- avx2, c, cpu-inference, gguf, inference-engine, linux, llm, local-llm, mixture-of-experts, moe, nvme, quantization, qwen3, simd, speculative-decoding, vulkan
- Fetched at
- Sept. 24, 2026, 5:02 p.m.
- Updated at
- Sept. 24, 2026, 5:02 p.m.
Enrichment
- Theme
- voice AI agents and infrastructure
- Vertical
- Horizontal
- Function
- Model & infra
- Audience
- Developer
- AI stance
- AI-native
- Project type
- Hobby / open-source project
- Normalized one-liner
- c-based mixture-of-experts inference engine streaming from nvme
- Manually corrected
- False
Could you build this?
No Streaming MoE model weights dynamically from NVMe drives in pure C requires advanced systems programming, custom memory-mapped I/O (io_uring/direct I/O), and low-level matrix math kernels.
What it would actually take: The architecture requires a custom inference engine in C utilizing asynchronous Linux I/O (`io_uring`) or DirectStorage to page gigabytes of expert weights off NVMe in real time matching token generation latencies. The hard engineering problems are cache replacement algorithms, latency hiding via prefetching predicted expert activation paths, and hand-optimized SIMD/GEMM routines. This demands high-performance computing (HPC) and deep systems programming expertise.
Competitors
Other products that read as similar to this one — 1097 launches clear the similarity bar, closest 8 shown.
Attention rank: #427 of 1098 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).
Launched 325 days after the earliest competitor.
- Open-source AMDGCN kernels for optimizing LLM inference · hn · 2026-08-25 · 5 upvotes · similarity 0.51
- OpenSourceJev · github · 2026-09-19 · 33 upvotes · similarity 0.51
- Tiny-vLLM · hn · 2026-05-29 · 205 upvotes · similarity 0.51
- voxweave · github · 2026-09-09 · 32 upvotes · similarity 0.51
- awesome-tts-architectures · github · 2026-09-13 · 15 upvotes · similarity 0.50
- jevcore · github · 2026-09-20 · 20 upvotes · similarity 0.50
- orukeet · github · 2026-09-09 · 40 upvotes · similarity 0.50
- LLM2Jev · github · 2026-09-19 · 248 upvotes · similarity 0.50
Other launches for this product
Same idea, different domain
Nobody's really built a model & infra tool for Fintech yet.