Qwen3.5-35B-A3B API Prices Jump 291%: Input Costs Triple Overnight

What Changed

Qwen has raised API pricing on Qwen3.5-35B-A3B across the board. Input token costs jumped from $0.08 to $0.313 per 1M tokens — that's a +291% increase. Output pricing moved more modestly but still meaningfully, from $0.75 to $1.25 per 1M tokens.

Does This Matter for Your Workload?

The answer depends heavily on your input-to-output ratio. Workloads that are input-heavy — long context retrieval, document processing, RAG pipelines — will feel this the most. A pipeline pushing 10M input tokens per day just went from costing roughly $0.80 to $3.13 on input alone. That's a real budget line item, not a rounding error.

Output-heavy use cases (long-form generation, code synthesis) see a smaller proportional hit — $0.75 to $1.25 is a 67% increase, significant but not the headline story here.

What to Do Next

Before accepting the new rate, it's worth checking whether the model's efficiency gains at this price point still beat alternatives. The MoE architecture (3B active parameters out of 35B total) was part of what made this model attractive at the old price — that calculus shifts at $0.313 input.

Check Qwen3.5-35B-A3B — live specs & price history to compare current pricing against similar MoE and dense models on the site before committing to new API contracts or scaling existing ones.

Bottom Line

A 291% input price increase is too large to absorb passively. Re-benchmark your cost-per-task against current alternatives — especially if input volume is your primary driver.

Ezra, Scout AI Team