One OpenAI-compatible gateway in front of Workers AI, freeAI gateways and SurfSense. Auth is optional — placeholder keys like ---------- are accepted.
Auth rules: no header needed. Bearer ---------- (any run of dashes), X-API-Key: ----------, or empty — all accepted as anonymous. A real token is only enforced if the SECRET_KEY secret is set on the Worker.
| Method | Path | Description |
|---|---|---|
| GET | / | Modern homepage + live playground |
| GET | /health | Worker + model registry status |
| GET | /v1/models · /models | OpenAI-style model list |
| GET | /v1/tier | Your current tier (anon / verified) |
| POST | /v1/chat | Simple chat — { model, messages, temperature, max_tokens } returns { ok, reply } |
| POST | /v1/chat/completions · /chat/completions | OpenAI-compatible completions (streaming supported) |
| POST | /v1/chat/{model-id} | Chat pinned to one model; voice models take { input } |
| POST | /v1/{system-prompt}/{model-id} | Smart prompting — hyphens become spaces |
| POST | /chat/v1/vision | Vision chat (image_url parts, qwen3.8-27b) |
| POST | /v1/image | Flux image generation — verified tier only |
| POST | /api/cmd | Terminal commands |
| GET | /verify | Discord verification flow |
| ID | Name | Vendor | Provider | Context | Parameters |
|---|---|---|---|---|---|
dg/gpt-oss | GPT-OSS | OpenAI | cloudflare | 128,000 tokens | 120B total · 12B active (MoE) |
dg/glm | GLM 4.7 Flash | Z.ai | cloudflare | 128,000 tokens | 9B (dense) |
dg/llama-4-scout | Llama 4 Scout | Meta | cloudflare | 128,000 tokens | 17B total (MoE) |
dg/deepseek | DeepSeek R1 Distill | DeepSeek | cloudflare | 128,000 tokens | 32B (dense) |
dg/gemma | Gemma-4-26B-a4b-it | cloudflare | 128,000 tokens | 26B (dense) | |
dg/qwen3.8-27b | Qwen3.8-27B | Alibaba | cloudflare | 128,000 tokens | 27B (dense) |
dg/flux-2-klein-9b | Flux-2-klein-9b | Black Forest Labs | cloudflare | Image generation | 9B (distilled) |
dg/qwen-coder | Qwen 2.5 Coder | Qwen | cloudflare | 32,768 tokens | 32B (dense) |
dg/agent | Agent | DG Develops | cloudflare | 32,768 tokens | 32B (dense) |
dg/dg-voice-1 | DG Voice 1.0 | DG Develops | cloudflare | Text Input | MeloTTS |
dg/dg-voice-2 | DG Voice 2.0 | DG Develops | cloudflare | Audio Input | Whisper + Llama 3.1 + MeloTTS |
dg/claude-sonnet-3.5 | Claude Sonnet 3.5 | Anthropic | freeaixyz | Long context | Advanced reasoning |
dg/gemini-3.6-flash | Gemini 3.6 Flash | Google DeepMind | freeaixyz | Long context | Fast inference |
dg/gemini-2.5-pro | Gemini 2.5 Pro | Google DeepMind | freeaixyz | Long context | General intelligence |
dg/gpt-5.4-mini | GPT 5.4 Mini | OpenAI | surfsense | Long context | Lightweight reasoning |
dg/jolly-gen-rp | Jolly Gen RP | JollyGen Labs | freeaixyz | Long context | Creative dialogue |
dg/minimax-m2.7 | Minimax M2.7 | Minimax AI | freeaixyz | Long context | General intelligence |
dg/grok-4 | Grok 4 | xAI | freeaixyz | 256,000 tokens | Advanced reasoning |
Voice: POST /v1/chat/dg-voice-1 with {"input":"hello"} returns MP3. Image: POST /v1/image with {"prompt":"..."} returns PNG.
The path segment becomes the system prompt (hyphens become spaces). Safety restrictions are appended automatically.
| Tier | Burst | Daily | Neuron budget |
|---|---|---|---|
| anon | 8 / 10s | 60,000 | 5,000 |
| verified | 16 / 10s | 82,000 | 8,550 |
Verify at /verify (Discord) for higher limits + image generation. 1 neuron is about 16 tokens.
check -m models · flag · average · owner -log in/out · ban -ip · unban -ip · give verify -ip · healthcheck
NVIDIA secret)