Animica — Python Cloud + pay-per-call x402 APIs for AI agents

# Animica.dev — free AI, Python cloud and coding agent for developers and AI agents > animica.dev is the developer platform of the Animica proof-of-work L1: a free, keyless, > OpenAI-compatible inference API, a Python function cloud that pays deployers in ANM, > a free GitHub-connected coding agent, GPU media studios — all served by the > Animica miner network — and Animica Health, verifiable healthcare AI infrastructure > with cryptographic execution receipts anyone can check. Current release: Animica 11.1.1 (2026-08-24). No signup and no API key are needed for the free AI API, the media studios or the coding agent. Rate limit on `/v1`: 30 requests/minute per IP (burst 12); over-limit requests get HTTP 429. Capacity comes from volunteer devices — phones, browsers and GPUs — not a datacenter. Be explicit about what that means before you send work: - Latency: a completion typically returns in 15–60 seconds; the measured tail reaches ~240 seconds. **Set your client timeout to at least 180s and prefer `"stream": true`**, or a default SDK timeout will abort a request that was going to succeed. - `GET /v1/models` carries `serving`, `latency_hint_s` and `context_length` per model. `serving: true` means a request WILL be answered (a worker is online, or the fallback path can answer). `serving: false` means don't send it — today that is the premium tier, which has no worker fielded. - Model substitution is disclosed, never silent: if the model that answers differs from the one you asked for, the response carries `x_animica_requested_model` and a `system_fingerprint` of the form `animica-substituted:<requested>-><served>`. The `model` field always names what actually produced the text. ## Free AI API (keyless, OpenAI-compatible) - [Near — natural-language local search](https://near.animica.org): describe the place you need in ordinary language ("a quiet coffee shop open after 8pm", "somewhere halfway between us for lunch", "a five-stop w