Spaces:
Running on Zero
Running on Zero
Commit History
-
feat(models): add Qwen3.8-27B Turbo Coder to the ZeroGPU catalog
AndrianBalanescuAdds the DavidAU TURBO NEO-CODER MAX build as a 27B option. The regular Q4_K_M quant is picked over the MTP variant on purpose: this integration never enables draft decoding, so MTP heads would only add weight. Also moves GGUF fetching out of the GPU lease. A cold 17 GB download far exceeds the 120 s budget, so a first call could time out and burn quota; weights are now fetched on CPU before gpu_tracked_call opens the lease. Fixes a test that touched the live Space: the remote MiniMax mock read .full_url unconditionally, so the string download leg raised, the route fell through to ZeroGPU, and the test depended on the daily quota.b08fc22
-
security+perf: enforce studio auth and cache singletons
AndrianBalanescuP0-1: add FLOW_API_KEY bearer-token middleware to studio_server.py (parity with app.py), with static/OPTIONS exemption and ?key= fallback. P1: cache DictionaryManager + ModelDispatcher via mtime/settings invalidation instead of per-request instantiation. Tests: update fixtures to send Authorization; add test_studio_auth_enforcement.py (7 tests). Closes P0-1 and P1 from audit.1a4837b
-
fix(hub): add model alias resolution and persist canonical ibrowse qa evidence
AndrianBalanescuf725c6a
-
test: pass FLOW_API_KEY bearer header in space_client fixture when set
AndrianBalanescuec2b860 -
feat(audio): purge Kokoro and enforce MiniMax Music 3 only policy for songs
AndrianBalanescu- Deleted 106 non-MiniMax song jsons, 14 fake audio files, and the 2026-09-03 fake album - Banned Kokoro repo-wide (replaced TTS with Edge-TTS ro-RO EmilNeural and F5-TTS references) - Enforced hard MiniMax Music 3 policy in ModelDispatcher.generate_music: non-MiniMax engines raise RuntimeError, zero fallback on unavailability - Updated daily_album_cron to enforce MiniMax segments, sha256 deduplication, and minimum 2 distinct segments per track - Replaced songEngine dropdowns and audio rack in static/index.html to MiniMax Music 3 only policy - Re-routed studio_server and media_creator song update/create to MiniMax-only dispatch with HTTP 400 on banned engines - Updated test suite with offline mocks to verify policy adherence (103 passed)8ad5a21
-
Fix ZeroGPU 120s cap, add async audio jobs, Gradio queue SSE client in dispatcher
AndrianBalanescu- Clamp ZeroGPU audio durations to 120s (platform ceiling) in generate_zerogpu_music and create_moldovan_song_handler - Extract _raw_generate_zerogpu_music to avoid nested @spaces.GPU leases - Add /v1/audio/jobs async background jobs API (POST /status /download) - /v1/audio/speech supports ?async=true or Prefer: respond-async - Add /info alias route for gradio_client compatibility - model_dispatcher: 3-tier resilient ZeroGPU audio (Gradio SSE queue -> async jobs -> sync) with local 808 synth fallback - daily_album_cron: 120s track duration, preserve real audio extension in album manifest - tests: unit coverage for async jobs endpointb98a720
-
fix(api): align frontend routes with media endpoints and dual route decorators
AndrianBalanescu4d8d6a7 -
feat(music-studio): wire Ion King Kong persona and voice cloner into AI Music Studio Pro
AndrianBalanescufbb5a31
-
feat(zerogpu): integrate petruhonk & Jackrong models, expand 24GB VRAM catalog, and add zerogpu-model-scout skill
AndrianBalanescu9f01856
-
feat(agentic): add offline test suite, agent contracts, and unified verification runner
AndrianBalanescud863d16