All pull requests

Pull requests

• cuda: run the DeepSeek V4.1 vision encoder

  #1315·francoisp opened Nov 3, 2026·

  ·

• CUDA: V4.1 routed experts on extra GPUs, parallel SSD expert reads, tested on 3090s with the 2bit quant

  #1313·francoisp opened Nov 2, 2026·

  ·

• ci: add a model-free pull request gate

  #1312·giuliowaitforitdavide opened Nov 1, 2026Contributor·

  ·
  1

• server: mask DeepSeek V4 (pre-4.1) chat syntax against control-token injection (stacked on #1288)

  #1310·Flor1an-B opened Nov 1, 2026·

  ·
  1

• Raise the server request body cap to 256 MiB

  #1309·ivanfioravanti opened Oct 31, 2026Contributor·

  ·
  4

• server: add -a/--alias to rename the OpenAI model id

  #1307·dineshr93 opened Oct 31, 2026·

  ·

• docs(jpeg): describe the malformed scan fixture in the fixture README

  #1306·giuliowaitforitdavide opened Oct 31, 2026Contributor·

  ·
  1

• server: accept OpenAI token-ID prompt arrays in /v1/completions

  #1304·potoior opened Oct 30, 2026·

  ·

• docs: add Jetson AGX Thor platform guide

  #1300·potoior opened Oct 29, 2026·2 tasks done·

  ·

• Accept published V4.1 DSpark metadata and Markov tensor names

  #1299·amittell opened Oct 28, 2026·

  ·
  1

• Add opt-in V4.1 decode timing breakdown (DS4_V41_STEP_TIMING, DS4_V41_LAYER_TIMING)

  #1298·amittell opened Oct 28, 2026·

  ·
  2

• server: render blank reasoning as absent (fixes full re-prefill after reasoning-free turns)

  #1296·timlawrenz opened Oct 28, 2026Contributor·

  ·
  2

• Agent: save exact token ids with KV sessions

  #1295·HDYS-TTBYS opened Oct 28, 2026Contributor·

  ·

• Add experimental native V4.1 weights for Metal SSD inference

  #1293·jomplox opened Oct 27, 2026·

  ·

• Add IQ1_M as a routed-expert tensor type

  #1291·yijunyu opened Oct 27, 2026·

  ·

• Metal: reuse f16 weight rows for 2-8 decode rows

  #1290·HDYS-TTBYS opened Oct 26, 2026Contributor·

  ·

• CUDA: 10x speedup for DeepSeek-V4 on RTX 3090 (via Antigravity AI agent) & statistical metrology (LMM/TOST)

  #1289·massimilianoporzio opened Oct 26, 2026·

  ·

• Server: mask GLM and Qwen chat renderers

  #1288·HDYS-TTBYS opened Oct 26, 2026Contributor·

  ·

• MiniMax M3.1 Flash: experimental Metal support with SSD streaming

  #1287·paradise-runner opened Oct 26, 2026Contributor·

  ·

• metal: reuse q8_0 weight loads for 2-8 decode rows

  #1286·HDYS-TTBYS opened Oct 26, 2026Contributor·

  ·

• metal: bf16 V4.1 index cache with tensor score path

  #1285·HDYS-TTBYS opened Oct 26, 2026Contributor·

  ·
  1

• Server: do not let chat content inject control tokens

  #1284·HDYS-TTBYS opened Oct 26, 2026Contributor·

  ·
  1

• v4.1: rewind sessions with images in the kept prefix

  #1283·gaineyllc opened Oct 26, 2026·

  ·
  1

• V4.1: pack compressed KV rows to 4 bits (bit-exact, a quarter of fp16)

  #1282·gaineyllc opened Oct 25, 2026·

  ·
  1

• server: a short live prefix no longer hides a longer disk-cache hit

  #1281·gaineyllc opened Oct 25, 2026·

  ·