Generate and edit with Klein, ZImage, Bonsai, HiDream, Krea 2, and Ideogram. The image commands support references, image-to-image generation, and API edits.

Train Klein and Krea 2 image adapters or text adapters from SFT JSONL. Preflight checks the plan before loading a model.

Turn one image into GLB, OBJ, PLY, or PBR voxels with MLX TRELLIS.2. TripoSR and InstantMesh create lower-detail drafts.

Generate with LTX 2.5, LTX 2.3, Wan, or MiniMax H3. Create synchronized audio-video, retake timed regions, Dub-It from a reference, or animate masked subjects with SCAIL-2.

Keep DreamX or Cosmos world state alive across camera-controlled transitions, or run a resident video session for repeated shots.

Recover metric depth, normals, point clouds, multi-view geometry, video depth, pose, and optical flow.

Analyze flood and fire tiles with TerraMind, encode Sentinel-1 and Sentinel-2 time series with TESSERA v2, or create multisensor spatial embeddings with OlmoEarth. Review all outputs as candidate features.

Generate songs with MiniMax Music 3 or ACE-Step. Create covers and retakes, run realtime sessions, analyze mixes, train adapters, export digital audio workstation bundles, or transcribe instrument-separated MIDI.

Extend speech or general audio to 48 kHz with AP-BWE or UniverSR. Generate audio with LTX 2.5, or use built-in RoFormer models to separate stems, reduce noise, and remove reverb.

Generate effects with Woosh or add motion-synced audio to video with MMAudio. Includes CLAP scoring and codec round trips.

Run Gemma 4, Qwen3.8, Ornith 1.5, Laguna, Nemotron, LFM, and GGUF models with reasoning, tools, structured JSON, LoRA, and memory-aware presets.

Reason across text, images, audio, and video with Nemotron 3 Nano Omni. For focused visual tasks, use Qwen3.8, Bonsai, Muse Glimmer, or LFM2.5-VL.

Embed text, images, or mixed records in one Qwen3-VL vector space. Build local retrieval-augmented generation with Qwen3, and anonymize personally identifiable information before content reaches another tool.

Synthesize or clone voices with Qwen3, transcribe with Parakeet or Qwen3, identify speakers with Sortformer diarization, and stream microphone transcription with reusable profiles.

Caption, inspect, ground, segment, and extract text from images with Falcon Perception, SAM 3.1, Infinity, and LightOn OCR.

Track prompted objects in video files or a live camera feed.

Detect and analyze faces locally, then produce identity embeddings for matching workflows without uploading source media.

Inspect the command catalog and runtime status, compare model tiers, estimate memory fit, download or remove model data, optimize supported checkpoints, save masked configuration, and run quality checks on installed models.

Serve OpenAI-compatible chat, embeddings, images, edits, speech, and transcription, then connect a local Open WebUI.

Browse companion tools, review each install command and release channel, approve an installation, and check plugin health from the CLI.

Author immutable workflow graphs, preflight them, then run the same plan locally, over SSH, through hosted Relay, or directly on another machine with durable material references.

Inspect local, SSH, Relay, and direct executors; pair another client to a Mac; watch, fetch, cancel, or retry runs; and preserve artifacts locally.

Validate content-addressed evaluation packs, compare models, adapters, and prompts, resume interrupted runs, and publish complete reports as immutable result records.

Use packaged offline cookbooks, guided machine setup, masked persistent configuration, and optional local coding-agent onboarding.