Qwen3.8 27B Decentered on QuickSilver Pro
Qwen3.8 27B Decentered runs on independent community Apple Silicon machines in the Share Compute pool. Community-served calls cost $0.238 input / $1.428 output per million tokens, 30% less than Qwen3.8 27B, with the same 1M-token context and 131K maximum output.
At a glance
Community Macs for coding agents and long documents at 30% lower token prices.
Pricing comparison ($/1M tokens)
| Provider | Input | Output | vs QSP |
|---|---|---|---|
| QuickSilver Pro | $0.238 | $1.428 | lowest-cost |
| Alibaba (qwen/qwen3.8-27b) | $0.425 | $2.55 | 44% lower |
| OpenAI (GPT-4o mini) | $0.15 | $0.60 | 138% more expensive |
When to use
Use the decentered variant when community processing suits your workload. It supports OpenAI-compatible Chat Completions and Responses, streaming and tools, and replies directly without hidden reasoning tokens. Independent community operators see your prompts and completions; read Privacy Policy Section 5 before sending content.
When to use something else
Choose Qwen3.8 27B or another model when requests must be processed only by cloud providers. If no community machine is free, or a node is offline, busy or times out, your request is served by Qwen3.8 27B on a cloud provider and billed at its full price: $0.34 input / $2.04 output per million tokens, with its cache rates ($0.068 read / $0.425 write).
Quickstart (curl)
curl https://api.quicksilverpro.io/v1/chat/completions \
-H "Authorization: Bearer $QSP_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen3.8-27b-decentered",
"messages": [{"role": "user", "content": "Hello!"}]
}'OpenAI-compatible. One-line migration via base_url.
FAQ
No. It replies directly without hidden reasoning tokens, like cloud-only Qwen3.8 27B.
A 1,000,000-token context window and up to 131,072 output tokens, OpenAI-compatible Chat Completions and Responses, streaming and tools. Set model="qwen3.8-27b-decentered" with your existing QSP base URL and key.
Use the decentered variant when community processing suits your workload. It supports OpenAI-compatible Chat Completions and Responses, streaming and tools, and replies directly without hidden reasoning tokens. Independent community operators see your prompts and completions; read Privacy Policy Section 5 before sending content.
The request falls back to cloud-only Qwen3.8 27B and is charged at that model's full price, including its cache rates. The 30% discount applies only to community-served calls. Cloud fallback calls do not pay contributors.