Chinese AI startup DeepSeek quietly released 'DeepSeek V4 Pro 0813'.
Models & Pricing | DeepSeek API Docs
https://api-docs.deepseek.com/quick_start/pricing/
DeepSeek's release of its latest model, the 'DeepSeek V4 Pro 0813,' has become a hot topic on social media forums and SNS.
Cline , an open-source AI coding agent, stated, 'DeepSeek V4 Pro 0813 has been quietly released. It has improved its Terminal Bench score by 15.8% from the preview model in April, and can be run at a cost of about 1/57th of Claude Fable 5 (which performed at a similar level). With 1.6 trillion parameters, 49 billion active parameters, and 1 million context windows, it is the best price-performance model of any AI model currently available on the market,' highlighting the excellent cost-performance of DeepSeek V4 Pro 0813.
DeepSeek silently V4-Pro 0813, up 15.8% on Terminal Bench from their April Preview model, with Fable 5 performance released at ~57x cheaper cost.
— Cline (@cline) August 12, 2026
1.6T param, 49B active, 1M context. This is the best price-to-performance model on the market right now.
Available in ClinePass now! pic.twitter.com/D9yas0umPn
@ChrisGPT , a writer for technology media outlets such as The Information and TechCrunch, posted: 'DeepSeek V4 Pro 0813 has been released. According to DeepSeek, the Terminal Bench 2.1 score has improved from 72.1% to 87.9%, the CyberGym score (a benchmark test that evaluates the cybersecurity capabilities of AI agents) has improved from 52.7% to 83.3%, and the DeepSWE score (a benchmark test that evaluates the performance of AI coding agents) has improved from 12.8% to 62.7%. The cost per million tokens is $0.435 (approximately 69 yen) for input and $0.87 (approximately 139 yen) for output.'
DeepSeek V4 Pro 0813 is out today too!
— Chris (@ChrisGPT) August 12, 2026
DeepSeek reports jumps from 72.1% to 87.9% on Terminal-Bench 2.1, 52.7% to 83.3% on CyberGym, and 12.8% to 62.7% on DeepSWE!!
It costs $0.435/M input and $0.87/M output with a 1M-token context window.
I don't like how every company report… pic.twitter.com/weE6cljPVr
The name 'DeepSeek-V4-Pro-0813' is also listed in the official DeepSeek API documentation. The usage fee for 'DeepSeek-V4-Pro-0813' is $0.003625 (0.58 yen) per 1 million input tokens (cache hit), $0.435 for 1 million input tokens (cache miss), and $0.87 per 1 million output tokens.
Models & Pricing | DeepSeek API Docs
https://api-docs.deepseek.com/quick_start/pricing/
Furthermore, the following are the results of a benchmark test performed on 'DeepSeek-V4-Pro-0813' to have the AI model draw a 'pelican riding a bicycle' in SVG format .
Related Posts:
in AI, Posted by logu_ii