Gonka Launches DeepSeek v4 Flash With 2,000 H100 Equivalent Hash Rate

Gonka deploys DeepSeek v4 Flash alongside Kimi and MiniMax models, reaching 2,000 H100 equivalent hash rate and processing 19.6B tokens in 24 hours.

Woofun AI reports that decentralized AI network Gonka has activated DeepSeek v4 Flash 0731, expanding its supported model lineup to include Kimi-K2.6 and MiniMax M2.7. The platform aggregates global GPU resources to provide OpenAI and Anthropic-compatible invocation services, with total network capacity nearing 2,000 H100 GPUs. In the last 24 hours, the system processed 19.6 billion AI tokens, with DeepSeek models accounting for over 12 billion, or more than 60% of the volume. Access is available via community brokers, and the router currently distributes $20 in free credits for mainstream model usage.