Issues

• splash-m1 1.6.0 crashes with "native loop did not answer status within 30 s" on Flint UD-Q3_K_XL under Goose tool-calling (M2 Max 32 GB)

  paperniuk/splash#56 · Filter by author vboyz103 opened on Nov 8, 2026
  3

• [M1 Max] Q5 GGUF: Native loop timeout on cached thinking requests (HTTP 503)

  paperniuk/splash#54 · Filter by author viktorwerner78 opened on Nov 7, 2026
  1

• [1.6.0-m1] Regression: Engine failed · native loop did not answer status within 30 s

  paperniuk/splash#51 · Filter by author ghisguth opened on Nov 6, 2026
  10

• Vision analysis keeps timing out

  paperniuk/splash#50 · Filter by author Rdx-ai-art opened on Nov 6, 2026
  5

• M1 Ultra 48-core: Qwen3.8-Flash-Next IQ3_S - 51 t/s @1K, 34 t/s @192K, SSD-streamed PLE table costs no speed

  paperniuk/splash#47 · Filter by author dustinogle opened on Nov 4, 2026
  1

• Request : How to add this splash m1 to LM Studio backend? ( the official splash is already supported)

  paperniuk/splash#41 · Filter by author Rdx-ai-art opened on Nov 1, 2026

• Ternary Bonsai 2 PQ2_0 fails at maximum_prefill on M1 Pro

  paperniuk/splash#22 · Filter by author arsold2002-cyber opened on Oct 21, 2026
  2

• Serve from the prepared cache when the GGUF source is gone (prepared weights as SSOT)

  paperniuk/splash#8 · Filter by author CarloArpini opened on Oct 10, 2026
  2

• Inconsistent Prefill and cache performance

  paperniuk/splash#7 · Filter by author Rdx-ai-art opened on Oct 9, 2026
  5

• NOT AN ISSUE. just wanna know gguf vs mlx performance

  paperniuk/splash#4 · Filter by author Rdx-ai-art opened on Sep 30, 2026
  2

• Broken cache?

  paperniuk/splash#3 · Filter by author brechtm opened on Sep 26, 2026
  8