Open-source platform for 3D AI agents. Turn text or a photo into a rigged, animated GLB avatar, give it an LLM brain, memory and a wallet, and embed it anywhere with one web component. Remote MCP server, x402 payments, Three.js, browser-native.
JavaScript 3d-generationai-agentsanimationavatardigital-humangenerative-aiglbgltfimage-to-3dllmmcpmcp-servermodel-context-protocolsolanatext-to-3dthreejsweb-componentswebglwebxrx402
docs(llm): document the free provider lanes and pin their env names
nirholas96c0be4bb4c2
diff --git a/.env.example b/.env.exampleindex 98991fdf6..8a92be6c9 100644--- a/.env.example+++ b/.env.example@@ -136,6 +136,12 @@ ZAI_API_KEY=# https://siliconflow.comSILICONFLOW_API_KEY=+# LLM7.io: gemini-3.1-flash-lite on an OpenAI-compatible endpoint, about+# 30 req/min per IP. This was a keyless rung until llm7.io retired its+# anonymous tier; unauthenticated calls now answer 401 invalid_api_key, so the+# rung is skipped without this key. Free key at https://dash.llm7.io/#/api-keys+LLM7_API_KEY=+# Cloudflare Workers AI (free tier, 10,000 neurons/day): 70B-class Llama at# the edge. The rung needs BOTH values; the OpenAI-compatible URL embeds the# account id. Create a Workers AI scoped token at https://dash.cloudflare.comdiff --git a/docs/free-llm-providers.md b/docs/free-llm-providers.mdindex 5c57aa0fe..2947d80ec 100644--- a/docs/free-llm-providers.md+++ b/docs/free-llm-providers.md@@ -6,7 +6,7 @@ The chain lives in [`api/_lib/llm.js`](../api/_lib/llm.js) (`providerChain()` +## The chain, in order-Free rungs run first, 70B-class models before smaller ones. A keyed rung is skipped when its env var is unset; the three keyless rungs are always present, so the chain can never be empty.+Free rungs run first, 70B-class models before smaller ones. A keyed rung is skipped when its env var is unset; the two keyless rungs (OVH, Pollinations) are always present, so the chain can never be empty.| # | Rung | Model | Key | Free tier ||---|------|-------|-----|-----------|@@ -22,7 +22,7 @@ Free rungs run first, 70B-class models before smaller ones. A keyed rung is skip| 10 | Gemini (AI Studio) | `gemini-2.5-flash-lite` | `GEMINI_API_KEY` | AI Studio free tier, no card. [aistudio.google.com](https://aistudio.google.com) || 11 | Vertex Gemini | `google/gemini-2.5-flash` | GCP service account | The reliability anchor: billed to platform GCP credits, no third-party quota. || 12 | Pollinations | `openai-fast` | none (keyless) | Anonymous, always on. |-| 13 | LLM7.io | `gemini-3.1-flash-lite` | none (keyless) | Anonymous, about 30 req/min per IP. |+| 13 | LLM7.io | `gemini-3.1-flash-lite` | `LLM7_API_KEY` | About 30 req/min per IP. Was keyless until llm7.io retired its anonymous tier; unauthenticated calls answer 401 `invalid_api_key` (measured 2026-09-02), so the rung is skipped without a key. Free key at https://dash.llm7.io/#/api-keys || 14 | SiliconFlow | `Qwen/Qwen3-8B` | `SILICONFLOW_API_KEY` | Free-tier small model, own quota pool. [siliconflow.com](https://siliconflow.com) || 15 | Groq instant | `openai/gpt-oss-20b` | `GROQ_API_KEY` | Separate per-model quota from rung 1. |diff --git a/docs/ops/llm-lanes.md b/docs/ops/llm-lanes.mdindex 3d2a50bc1..967e4c81b 100644--- a/docs/ops/llm-lanes.md+++ b/docs/ops/llm-lanes.md@@ -31,7 +31,7 @@ production are routinely dead and a chain that depends on them fails:| 10 | Gemini Flash-Lite (AI Studio) | `GEMINI_API_KEY` | free | not configured in prod || 11 | **Vertex Gemini Flash** | GCP service account | GCP credits | **serving**, the reliability anchor || 12 | Pollinations `openai-fast` | none (keyless) | free | **serving** (200, ~3.7s, 2026-08-02) |-| 13 | LLM7.io `gemini-3.1-flash-lite` | none (keyless) | free | added 2026-08-05, the third keyless rung |+| 13 | LLM7.io `gemini-3.1-flash-lite` | `LLM7_API_KEY` | free | **dead without a key**: llm7.io retired the anonymous tier it was added on, and every unauthenticated call answers 401 `invalid_api_key` (measured 2026-09-02, the `unused` token its docs used to accept included). Key-gated since, so a deployment without the key skips it. Free key at https://dash.llm7.io/#/api-keys || 14 | SiliconFlow `Qwen/Qwen3-8B` | `SILICONFLOW_API_KEY` | free | added 2026-08-05; skipped when the key is unset || 15 | Groq `llama-3.1-8b-instant` | `GROQ_API_KEY` | free | serving (separate per-model quota) || 16 | Vertex Claude | GCP service account + `VERTEX_CLAUDE_ENABLED=1` | GCP credits | **off and unentitled** (see below) |@@ -42,7 +42,11 @@ production are routinely dead and a chain that depends on them fails:The 2026-08-05 widening (SambaNova, Mistral, Z.AI, Cloudflare, LLM7,SiliconFlow) added six independent free quota pools; each is documented with-its tier limits in `docs/free-llm-providers.md`.+its tier limits in `docs/free-llm-providers.md`. LLM7 joined that round as a+keyless rung and is no longer one (row 13), so the keyless floor a zero-env+deployment falls to is OVH and Pollinations, two rungs rather than three. Both+answered 429 from this workspace's shared egress IP on 2026-09-02, so treat the+keyless floor as best-effort: it is a real lane, not a substitute for a key.Rungs 1 to 15 are not a degradation path any more. They are production. Everyone of them is covered by a transport-level failover testdiff --git a/tests/api/llm.test.js b/tests/api/llm.test.jsindex c959e5d79..69b10d1ee 100644--- a/tests/api/llm.test.js+++ b/tests/api/llm.test.js@@ -603,7 +603,7 @@ describe('llmComplete — failure modes', () => {},text: async () => 'not json',};- // Every reachable lane (groq + the three keyless rungs) returns garbage.+ // Every reachable lane (groq + the two keyless rungs) returns garbage.installFetch({ [GROQ_HOST]: badBody, [OVH_HOST]: badBody, [POLLINATIONS_HOST]: badBody, [LLM7_HOST]: badBody });await expect(llm.llmComplete({ system: 's', user: 'u' })).rejects.toMatchObject({status: 502,