ODS V3 Pre-Release: Public testing and refinement ahead of the official V3 launch. Turn your PC, Mac, or Linux box into a private AI server.
Python ai-agentsamdcomfyuidockerllama-cppllmlocal-ain8nnvidiaopen-webuiragself-hostedspeech-to-textstrix-halotext-to-speechworkflow-automation
Commits on main
fix(installer): refresh stale Hermes WSL file aliases (#8311)
a1073a2 Mike Bradley committed 21 days ago
Merge pull request #8314 from Osmantic/codex/fix-secret-scan-history-20261028
c3a0dab Mike Bradley committed 21 days ago
Merge pull request #8313 from Osmantic/elf-of-doom-shasan-leveling
24b5f75 Mike Bradley committed 22 days ago
Elf of Doom: HUD after Continue, steady leveling and Shasan
9a6b3ca Mike Bradley committed 22 days ago
fix(pixel): keep execution guidance out of owner replies (#8296)
5d4e5ad Mike Bradley committed 22 days ago
fix(pixel): carry verified delivery corrections into native history (#8312)
b67f8fc Mike Bradley committed 22 days ago
fix(installer): say "existing install" only when there is one (#8308)
c330588 Mike Bradley committed 22 days ago
Merge pull request #8310 from Osmantic/feat/elf-of-doom-endless-gabriel-blessing
0cf0123 Mike Bradley committed 22 days ago
fix(installer): name who holds a busy port, and every busy Dashboard port at once (#8307)
da3eaa7 Mike Bradley committed 22 days ago
fix(installer): give vLLM only a GPU with its share free on a fresh install (#8306)
bffd606 Mike Bradley committed 22 days ago
fix(nvidia): limit llama-server to its assigned GPUs with CUDA_VISIBLE_DEVICES (#8305)
5b3b2dd Mike Bradley committed 22 days ago
feat(vllm): run the RTX PRO 6000 class on vLLM, at half the GPU (#8303)
07e5516 Mike Bradley committed 22 days ago
fix(dashboard-api): don't make readiness wait for a switched-off Open WebUI (#8301)
3238b1c Mike Bradley committed 22 days ago
feat(installer): run the vLLM engine where a vLLM model fits one GPU (#8299)
666056e Mike Bradley committed 22 days ago
feat(vllm): run an LLM_BACKEND=vllm install end to end (#8298)
d06bb84 Mike Bradley committed 22 days ago
test(macos): keep every real stat field in the stop-witness cleanup test (#8300)
8b832b7 Mike Bradley committed 22 days ago
feat(pixel): let preview inspection start on another page of the snapshot (#8297)
eb83e43 Mike Bradley committed 23 days ago
perf(catalog): skip GLM's FlashInfer autotune, which bought nothing (#8295)
e6bf5ea Mike Bradley committed 23 days ago
fix(vllm): keep every kernel cache in the mounted cache folder (#8294)
4e7e511 Mike Bradley committed 23 days ago
fix(macos): accept LLAMA_REASONING in any case so llama-server starts (#8220)
c1a7920 Satyam Singh committed 23 days ago
fix(installer): wait for transient source hold lock contention (#8292)
6897732 Mike Bradley committed 23 days ago
feat(models): add vLLM catalog entries, selection and verified download (#8289)
68c1c9c Mike Bradley committed 23 days ago
fix(pixel): keep scratch and test pages out of published sites (#8291)
781d62d Mike Bradley committed 23 days ago
feat(compose): add the vLLM engine service for LLM_BACKEND=vllm (#8290)
9ed172f Mike Bradley committed 23 days ago
fix(llama): accept Docker reasoning modes in any case (#8279)
3221546 Mike Bradley committed 23 days ago
feat(models): add the vllm backend kind and its router endpoint (#8288)
a2c9c62 Mike Bradley committed 23 days ago
fix(pixel): reproduce a reported bug before fixing it (#8287)
590bee0 Mike Bradley committed 23 days ago
fix(portal): label unavailable inline images clearly (#8286)
63cba56 Mike Bradley committed 23 days ago