Qwen 27B models see opposing price moves on OpenRouter
OpenRouter updated pricing for two Qwen 27B models on November 1, with Qwen3.6-27B input costs rising 40.6% and output costs falling 15.6%, while Qwen3.8-27B input costs dropped 73.3% and output costs rose 22.7% compared to October 31.
Qwen3.6-27B pricing
Qwen3.6-27B input price increased from $0.32 to $0.45 per million tokens, a 40.6% rise. Output price decreased from $3.20 to $2.70 per million tokens, a 15.6% cut. The model offers a 262,144-token context window and was released on April 27, 2026.
Qwen3.8-27B pricing
Qwen3.8-27B input price fell from $0.15 to $0.04 per million tokens, a 73.3% decrease. Output price rose from $1.875 to $2.30 per million tokens, a 22.7% increase. This model provides a 1,000,000-token context window and was released on August 14, 2026.
Model specifications
According to Qwen, Qwen3.6-27B is a causal language model with a vision encoder containing 27B parameters, 64 layers, hidden dimension of 5120, and a token embedding size of 248,320. The architecture uses a hybrid layout of 16 blocks with 3 Gated DeltaNet layers followed by 1 Gated Attention layer per block. Context length is 262,144 tokens natively, extensible up to 1,010,000 tokens.
Key takeaways
- Qwen3.6-27B input pricing rose sharply while output pricing declined moderately.
- Qwen3.8-27B input pricing dropped dramatically while output pricing increased.
- Both 27B models remain available with open weights on OpenRouter.
Official announcement
Published by Qwen. Read it at the source.