Access AI inference, fine-tuning, and GPU rentals across DCDN Cloud's decentralized node network.
The GPU Marketplace provides three core services:
Use our API as a drop-in replacement for OpenAI. Just change the base URL:
curl https://dcdncloud.com/api/v1/gpu/inference \
-H "Authorization: Bearer YOUR_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"model": "dcdn-smart",
"messages": [{"role": "user", "content": "Hello!"}],
"max_tokens": 200
}'
| Model ID | Description | Pricing |
|---|---|---|
dcdn-smart | Claude Sonnet 4.6 — best price/performance | $3 / $15 per 1M tokens |
dcdn-fast | Fast inference, lower latency | $0.25 / $1.25 per 1M tokens |
dcdn-reasoning | Deep reasoning model | $3 / $15 per 1M tokens |
gpt-4o | OpenAI GPT-4o | $2.50 / $10 per 1M tokens |
gpt-4o-mini | OpenAI GPT-4o Mini | $0.15 / $0.60 per 1M tokens |
claude-* | Anthropic Claude models | Varies by model |
gemini-* | Google Gemini models | Varies by model |
deepseek-* | DeepSeek models | Varies by model |
meta-llama/* | Meta Llama models | Varies by model |
21 named models available, plus any OpenRouter model ID works as passthrough.
💡 Pricing uses prefix matching — the most specific model prefix wins. Input and output tokens are priced separately.
curl https://dcdncloud.com/api/v1/gpu/inference/stream \
-H "Authorization: Bearer YOUR_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"model": "dcdn-fast",
"messages": [{"role": "user", "content": "Write a poem"}],
"stream": true
}'
curl -X POST https://dcdncloud.com/api/v1/gpu/inference/batch \
-H "Authorization: Bearer YOUR_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"model": "dcdn-smart",
"prompts": ["Hello", "Summarize quantum computing", "Translate to French: Good morning"],
"max_tokens": 200
}'
curl -X POST https://dcdncloud.com/api/v1/gpu/training/create \
-H "Authorization: Bearer YOUR_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"model": "meta-llama/Llama-3-8B",
"dataset_url": "https://example.com/data.jsonl",
"epochs": 3,
"learning_rate": 2e-5
}'
# List all training jobs
curl https://dcdncloud.com/api/v1/gpu/training/jobs \
-H "Authorization: Bearer YOUR_TOKEN"
# Get specific job
curl https://dcdncloud.com/api/v1/gpu/training/jobs/{job_id} \
-H "Authorization: Bearer YOUR_TOKEN"
curl -X DELETE https://dcdncloud.com/api/v1/gpu/training/jobs/{job_id} \
-H "Authorization: Bearer YOUR_TOKEN"
Rent bare GPU instances by the hour — similar to Vast.ai or RunPod.
curl -X POST https://dcdncloud.com/api/v1/gpu/rent \
-H "Authorization: Bearer YOUR_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"node_id": "DCDN-SA-c47e",
"hours": 4
}'
curl https://dcdncloud.com/api/v1/gpu/rentals \ -H "Authorization: Bearer YOUR_TOKEN"
curl -X DELETE https://dcdncloud.com/api/v1/gpu/rentals/{id} \
-H "Authorization: Bearer YOUR_TOKEN"
GPU rental prices are set by node operators. The platform does not add markups. Revenue split:
The GPU Marketplace dashboard includes:
AI agents can pay for inference per-request using the x402 protocol (USDC on Base chain) — no API key needed.