Qwen3.6-35B-A3B for long context
Qwen3.6-35B-A3B is a 35B-parameter MoE with only 3B active per token and a 262K context window. The MoE lets it run at a 3B-dense cost while keeping a 35B knowledge base - ideal for RAG and long-document workflows. It costs $0.112 input / $0.80 output per 1M tokens.
Why it's a fit for RAG
262K context: Fits a 500-page PDF or 200 code files into a single prompt. No need for aggressive chunking if the retrieved corpus fits; single-shot RAG simplifies your pipeline.
Low input cost: the current catalog rate is $0.112 per 1M input tokens. Frontier models cost many times more for the same prompt.
MoE speed: Only 3B parameters are active per token, so inference speed is closer to a 3B dense model than a 35B dense one. For long-input workflows, this shows up as noticeably lower per-request latency.
RAG pipeline pattern
Simple single-shot: if retrieved context fits in 262K tokens, skip reranking and hierarchical summarization - feed everything to Qwen3.6-35B-A3B in one call. Lower pipeline complexity, lower latency.
With retrieval: embed -> top-K retrieve -> concat into a 50-100K token prompt -> Qwen3.6-35B-A3B answer. Input-cost economics favor longer top-K because input tokens cost little.
Summarize-then-answer: for corpora larger than the model context, first summarize by section, then answer on the summaries.
Quickstart code
from openai import OpenAI
client = OpenAI(
base_url="https://api.quicksilverpro.io/v1",
api_key="sk-qsp-...",
)
document = open("annual-report.txt").read()
resp = client.chat.completions.create(
model="qwen3.6-35b",
messages=[
{"role": "system", "content": "Answer using only the provided document."},
{"role": "user", "content": f"Document:
{document}
Question: What was free cash flow in Q3?"},
],
max_tokens=500,
)
print(resp.choices[0].message.content)FAQ
Yes. 262K tokens is the catalog limit. Long-context recall can still degrade near a model's hard limit, so critical retrieval should combine vector search with prompts that put the most relevant chunks first.
Mixture-of-Experts routes each token through only a subset of the model's parameters. Qwen3.6-35B-A3B has 35B total parameters but activates only 3B per token. This makes long-context workloads a particularly good fit.
The model supports reasoning mode. QuickSilver Pro suppresses reasoning by default to keep output concise and predictable. Pass reasoning.enabled=true to opt back in; reasoning tokens count as output.
Its catalog capabilities are text input, text output, streaming, reasoning, tool calling and JSON Schema, exposed through Chat Completions and Responses. Benchmark complex multi-tool loops on your own workload before committing.