Anuncio

SRExpert

Deja de saltar entre 6 herramientas de Kubernetes.

Monitoreo, alertas, seguridad, despliegues Helm y operaciones con IA en una sola plataforma.

Probar 30 días gratis

Qwen

Qwen3.5 397B A17B cuts input price 25% on OpenRouter

460 palabras

Por ahora esta nota solo está disponible en inglés.

Qwen cut the OpenRouter price of Qwen: Qwen3.5 397B A17B on 2026-11-17, to $0.45 per million input tokens and $3.00 per million output tokens, from $0.60 and $3.60 the day before, a 25.0% drop on input and 16.7% on output.

Precio en OpenRouter, por 1M de tokens

Medido el 17 de noviembre de 2026

Qwen: Qwen3.5 397B A17B
Entrada /M tokens
0,45 US$
Salida /M tokens
3,00 US$

Cambios de precio en los últimos 90 días

  1. entrada 0,60 US$ → 0,45 US$, salida 3,60 US$ → 3,00 US$
  2. entrada 0,50 US$ → 0,60 US$, salida 3,60 US$ → 3,60 US$
  3. entrada 0,39 US$ → 0,50 US$, salida 2,34 US$ → 3,60 US$
  4. entrada 0,50 US$ → 0,39 US$, salida 3,60 US$ → 2,34 US$
  5. entrada 0,55 US$ → 0,50 US$, salida 3,50 US$ → 3,60 US$
  6. entrada 0,45 US$ → 0,55 US$, salida 3,00 US$ → 3,50 US$

Historial completo de precios

Where today's price sits

The model's OpenRouter record starts on 2026-08-22 and covers 88 days, during which the price has changed 12 times, today's included. The previous change was on 2026-11-10, when input went to $0.60 per million tokens and output to $3.60.

The lowest input price in the record is $0.39 per million tokens, first seen on 2026-08-22; the highest is $0.60, first seen on 2026-11-10. Today's $0.45 is a return to a level the record already shows: input was $0.50 per million tokens from 2026-10-15, $0.39 from 2026-10-21, $0.50 again from 2026-10-23 and $0.60 from 2026-11-10. Output at $3.00 per million tokens is below the $3.60 that started on 2026-10-15 and returned on 2026-10-23 and 2026-11-10, and above the $2.34 set on 2026-10-21.

Next to other Qwen models

Qwen: Qwen3.5 397B A17B carries a 262,144-token context and open weights. Among Qwen models on OpenRouter today, Qwen: Qwen3.6 27B also lists a 262,144-token context, at $0.32 per million input tokens and $3.20 per million output tokens. Qwen: Qwen3 235B A22B Thinking 2507 lists $0.45 per million input tokens and $3.50 per million output tokens over 128,000 tokens.

Qwen2.5 72B Instruct is listed at $0.36 per million input tokens and $0.40 per million output tokens over 32,768 tokens. The closed-weight Qwen: Qwen3.6 Plus is at $0.325 per million input tokens and $1.95 per million output tokens over 1,000,000 tokens, and Qwen: Qwen3.7 Plus at $0.32 and $1.28 per million tokens over the same context.

What Qwen published

Qwen's announcement, titled Qwen/Qwen3.5-397B-A17B, lists 397B parameters in total and 17B activated, a hidden dimension of 4096, a token embedding of 248320 (Padded), 60 layers, 512 experts, 10 routed experts plus 1 shared activated expert, and an expert intermediate dimension of 1024. Qwen says the model has a context length of 262,144 natively, extensible up to 1,010,000 tokens, and supports 201 languages and dialects.

According to Qwen, the weights are compatible with Hugging Face Transformers, vLLM, SGLang and KTransformers, and the official Qwen API service is provided by Alibaba Cloud Model Studio. Qwen says Qwen3.5-Plus is the hosted version corresponding to Qwen3.5-397B-A17B with 1M context length by default.

  • Input fell 25.0% to $0.45 per million tokens and output 16.7% to $3.00 per million tokens in one day.
  • Today's input price matches a level already seen in the 88-day record; the record's low is $0.39 and its high $0.60.
  • Qwen lists the model at 397B total and 17B activated parameters with a native 262,144-token context.
Qwen3.5 397B A17B cuts input price 25% on OpenRouter — overfeed.news