Qwen3.8 27B Drops Input Price to $0.40 and Output to $3.00 per 1M Tokens

What Changed

Qwen has quietly trimmed the API pricing on Qwen3.8 27B. Input tokens dropped from $0.45 to $0.40 per 1M tokens, and output tokens fell from $3.20 to $3.00 per 1M tokens — an overall cut of around -11%.

Does It Move the Needle?

For input-heavy workloads — think document processing, long-context retrieval, or RAG pipelines — the saving is modest: $0.05 per million tokens. At serious scale (say, 500M input tokens/month), that's $25 back in your pocket. Not transformative, but not nothing either.

The output reduction is proportionally more meaningful. Output tokens are where costs compound fastest in chat, code generation, and summarisation tasks. Dropping from $3.20 to $3.00 per 1M tokens gives you a small but real buffer if your app is output-heavy.

Who Should Care

• Developers running mid-size model comparisons: Qwen3.8 27B sits in a competitive bracket. This cut keeps it in contention against similarly-sized alternatives without requiring a model swap.
• Cost-optimisers already on Qwen3.8 27B: No action needed — your bills just got marginally smaller.
• Teams evaluating the model fresh: The new pricing makes a pilot run cheaper, lowering the cost of a proper evaluation.

Bottom Line

This isn't a dramatic repricing event, but it's a clean, no-strings reduction across both token types. If you're already factoring Qwen3.8 27B into a shortlist, the numbers just improved slightly in its favour. Check current rates and compare against alternatives on the Qwen3.8 27B — live specs & price history page before committing to a workload.

Ezra, Scout AI Team