mlx-serve

mlx-serve runs MLX models on this Mac behind one HTTP port: OpenAI, Anthropic and Ollama compatible chat, plus native endpoints for images, speech, music, sound effects, video, 3D and decisions. Everything below is plain HTTP + JSON, so any language works.

Connect

• Base URL: $MLX_SERVE_URL when set, else http://127.0.0.1:11234. Make it a config value or env var in the code you write, never a hardcoded LAN IP.
• Auth: none from this Mac. A server started with --api-key wants Authorization: Bearer <key> from other machines; a 401 means that.
• CORS is open (Access-Control-Allow-Origin: *): a browser game can call it directly.
• Up? curl -s "$MLX_SERVE_URL/health"

Pick models from /v1/models, never guess ids

curl -s "${MLX_SERVE_URL:-http://127.0.0.1:11234}/v1/models" \
  | jq -r '.data[] | "\(.id)\t\(.capabilities | join(","))\t\(.state)"'

Each row has id (like org/name), capabilities, state (ready, unloaded, remote) and context_length. Choose by capability:

Read the linked file (next to this one) before writing client code for that endpoint. If no model has the capability the user needs, say so and tell them to download one in the MLX Core app (Models). Do not invent an id.

Loading and memory

• Any request names its model with "model": "<id>". An unloaded model loads on the first request, which can take seconds to minutes. POST /v1/load-model {"model": "<id>"} pre-warms one; POST /v1/unload-model {"model": "<id>"} frees it. While a load runs, GET /v1/load-progress lists each loading model's phase, done/total and eta_s.
• All models share one GPU memory budget. A 503 out_of_memory means unload something first (typically the media model when you are done with it).
• Generation runs on one inference thread: a 3-minute video render stalls chat replies for 3 minutes. Decisions and embeddings are milliseconds.

Films and music videos

Making a video (music video, trailer, release or story film)? Read film.md first: a YuE2 song is the spine, gen.mjs makes the pictures, clips and sound effects, and film.mjs (Node 18+) times every cut to the sung words and renders it.

Rules for apps and games

• Building a 3D game or scene? Read games.md first: movement, foliage and lighting rules, and gen.mjs (next to this file, Node 18+), which generates and post-processes every asset from one assets.json (node gen.mjs help).
• Generate media AHEAD of time or in the background, never on the frame or input path. Cache results on disk keyed by model + prompt + seed + size, and ship the cache (or regenerate lazily) instead of calling on every run.
• Rough cost on Apple Silicon: image 5-60 s, speech about real time or faster, sound effects about a second, music 30 s to minutes, 3D 1-5 min, video minutes. Set client timeouts to match (10+ minutes for video), and keep the game playable while waiting.
• Send "seed" for reproducible assets.
• Long jobs: send "stream": true and read the SSE progress events to drive a progress bar. See media.md.
• Errors are JSON: {"error": {"message": "..."}} with a 4xx/5xx status. The message names the bad field. Show it; never retry a 400 unchanged.
• Keep the server's own URL and model ids in one config module so the user can swap models without touching game code.

Smoke test

BASE="${MLX_SERVE_URL:-http://127.0.0.1:11234}"
IMG=$(curl -s "$BASE/v1/models" | jq -r '[.data[] | select(.capabilities | index("image"))][0].id')
curl -s "$BASE/v1/images/generations" -H 'content-type: application/json' \
  -d "{\"model\":\"$IMG\",\"prompt\":\"pixel art treasure chest on a plain white background\",\"size\":\"512x512\",\"seed\":7}" \
  | jq -r '.data[0].b64_json' | base64 -d > chest.png