DeepSeek V4 Flash (Free)

DeepSeek V4 Flash hosted free by NVIDIA. 284B / 13B active MoE, 1M context, ~5x faster than V4 Pro. Best for chat, summarization, light reasoning. Weaker factual recall — pick V4 Pro for fact-heavy agentic loops

Code Examples

from blockrun_llm import LLMClient

client = LLMClient()  # Uses BLOCKRUN_WALLET_KEY (never sent to server)
response = client.chat("nvidia/deepseek-v4-flash", "Hello!")

Pricing

Payment