Skip to content
AI Model Radar

DeepSeek V4 Flash

deepseek-ai/DeepSeek-V4-Flash
Open Source · chat · open-weights
GA Alert me on changes
Intelligence
40.3
Context
1M
Max output
—
Input $/1M
$0.09
Output $/1M
$0.22
Modalities
text
Released
22 Apr 2026
Intelligence Index via Artificial Analysis · 0–100, higher is better
License: mit · deepseek-ai/DeepSeek-V4-Flash
AI summary
● machine-written

DeepSeek V4 Flash: 284B MoE model with 1M context window

DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model with 284B total parameters and 13B activated parameters, designed for fast inference and high-throughput workloads. It supports a 1M-token context window and is suited for coding assistants, chat systems, and agent workflows where responsiveness and cost efficiency are important.

What's new
  • Supports hybrid attention for efficient long-context processing
  • Reasoning efforts high and xhigh are supported; xhigh maps to max reasoning
  • Available through multiple inference providers including Novita, Fireworks AI, and DeepInfra
Best for
Coding assistantsChat systemsAgent workflowsHigh-throughput inference
Sources

Source: https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash