SPB Git forge
spboucher.ai ↗

spb/llm-api

Public
0commits 0branches 0releases
0 Bsize
maindefault branch
—last push

page 1

  1. Thu, Sep 10, 2026 5

    1. UI proxy: keep Content-Encoding so compressed Next.js responses render in browsers
      …
      Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
      Simon-Pierre Boucher committed 13 days ago · 1 file +2 −1
    2. Harvester: correct MLX weight estimates from HF metadata, flag installed mirrors; docs: managed-firewall WireGuard relay
      …
      Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
      Simon-Pierre Boucher committed 13 days ago · 3 files +14 −6
    3. Harvester: prefer the Apple Silicon runtime, MoE-aware KV fallback, size-constrained starter slots; README perf table
      …
      Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
      Simon-Pierre Boucher committed 13 days ago · 2 files +43 −21
    4. Harvester: skip repositories without weight files instead of aborting the scan
      …
      Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
      Simon-Pierre Boucher committed 13 days ago · 2 files +7 −1
    5. LLM API v0.1.0 — private OpenAI-compatible local model server for Apple Silicon
      …
      FastAPI gateway + SQLite registry, MLX-LM / mlx-vlm and llama.cpp workers, load-on-demand
with memory policy and eviction, OpenAI endpoints (chat/completions/embeddings/rerank, streaming),
management API, Model Harvester, benchmarks, API keys, Next.js 16 console, tests and docs.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
      Simon-Pierre Boucher committed 13 days ago · 70 files +13,830