NewsBite analysis · AI, Cloud, Startups
wccftech
DeepSeek Prices Its New V4-Pro-0813 Model At $0.87 Per 1 Million Output Tokens, As The High-Flying Chinese AI Lab Wows With Its Soaring Token Consumption
DeepSeek, the emergent Chinese AI lab, has rolled out its latest V4-Pro-0813 model aimed at undercutting competitors like OpenAI. Priced competitively, the model represents a strategic maneuver in an ongoing AI pricing war sparked by OpenAI. The announcement follows significant performance benchmarks and increased token consumption by DeepSeek, indicative of its rising dominance in the AI market.
Act nowConfidence: HighShort term horizon
Published 12 Aug 2026, 18:20 UTC3 min read
Visit original article ↗Lead image — wccftech
Summary
TL;DR
- DeepSeek's V4-Pro-0813 model priced at $0.87 per million output tokens
- Model performance parallels Anthropic's Opus 4.8 despite using fewer parameters
- DeepSeek has experienced a spike in demand amid an AI pricing war
- Token prices introduced to directly compete with OpenAI's discounted offerings
- Limited compute capacity raises concerns over model inference speeds
Interesting facts
- DeepSeek was second only to Anthropic in token consumption in July.
- The V4-Pro model's performance is expected to challenge established models despite lower parameters.
- DeepSeek's capacity is currently limited to 20,000 NVIDIA H100 GPUs.