Composite index published by LLM Stats. It is not a percentage of correct answers.
DeepSeek V4 Flash
Modelo de texto con herramientas, salida JSON y modos thinking y no-thinking.
Checked on 2026-09-08Add to a comparison We'll only use your verified email for this alert. No marketing.
LLM Stats reasoning index, not a measurement of our own.
LLM Stats coding index; review its methodology before reading into differences.
Standard uncached input price for short-context requests.
Standard output token price; reasoning tokens may be billed.
Published limit: total window for OpenAI/Claude, input limit for Gemini. It doesn't measure effective recall.
Output limit documented by the vendor.
Specs
- Contexto publicado
- 1M; salida máxima publicada: 384K
- Entrada
- Texto
- Identificador API
- deepseek-v4-flash
- Tarifa de referencia
- Pico, cache miss; fuera de pico 50%
- Versión consultada
- DeepSeek-V4-Flash-0731
Atlas · version 1 · Reviewed on 2026-09-08 · Not a real-time feed · Evidence d15f1a66e958
De
DeepSeek V4 Flash
Se muestra tarifa PICO sin caché. Horas pico: lunes a viernes 01–04 y 06–10 UTC. Fuera de pico cuesta la mitad. Las cifras 1M/384K conservan la notación publicada; los tokens de razonamiento consumen presupuesto.
Checked on 2026-09-08 · Editorial update