D
DeepSeek V4.1 Flash (batch)OPEN
DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture.
Los precios se actualizan automáticamente; se verifican a diario contra las tarifas oficiales de los proveedores.
- $0.17 por 1M de tokens, asumiendo 3 entradas por 1 salida la convención utilizada por Artificial Analysis
- más barato que el 64% de los 204 modelos open-weight rastreados aquí
- 3 de 15 modelos DeepSeek rastreados aquí, desde el más barato
Puntuaciones de referencia y rendimiento
Independent benchmark scores for DeepSeek V4.1 Flash (batch), measured by Artificial Analysis. Higher is better. Modo de medición: Max.
For the same budget, DeepSeek V4.1 Flash (batch) delivers 10 times more capability. Open weights also mean you can self-host it and pay nothing per token.
Los modelos en este nivel cuestan $2 por 1M de tokens (mediana de 19) — este cuesta $0.17.
Más barato y mejor en este índice: GLM 5.3 Flash (batch)
Acerca de este modelo
DeepSeek V4.1 Flash (batch) is an open-weight AI model by DeepSeek. You can download and self-host it for free; the prices below are hosted-API list prices, tracked daily, for when you prefer convenience over self-hosting.
Preguntas frecuentes
What is DeepSeek V4.1 Flash (batch)?
DeepSeek V4.1 Flash (batch) es un modelo de lenguaje de IA de DeepSeek. Es open-weight: puede descargarlo y ejecutarlo en su propio hardware, de forma gratuita. Obtiene 39.5 en el índice de inteligencia Artificial Analysis.
Is DeepSeek V4.1 Flash (batch) free?
The weights are free and open — you can self-host DeepSeek V4.1 Flash (batch) and pay nothing per token. If you prefer a hosted API, list prices are $0.11 per million input tokens and $0.34 per million output tokens.
What is DeepSeek V4.1 Flash (batch) good at?
Los benchmarks independientes de Artificial Analysis le dan Humanity's Last Exam 39.2%, Long Context Reasoning 84%, SciCode 51.9%.
How fast is DeepSeek V4.1 Flash (batch)?
Genera aproximadamente 226.5 tokens por segundo, con una mediana 0.84s de retraso antes del primer token. Medido de forma independiente por Artificial Analysis.
Can I self-host DeepSeek V4.1 Flash (batch)?
Yes. DeepSeek V4.1 Flash (batch) has open weights, so you can download it and run it on your own GPU or server with tools like Ollama, vLLM or llama.cpp — with no per-token cost.