Crypto
Home›Crypto›Market Structure›DeepSeek quietly updates V4 Pro pricing with a new 081…
DeepSeek quietly updates V4 Pro pricing with a new 0813 build
The finished DeepSeek-V4-Pro-0813 model is now listed on its API pricing page with the same token-cost rates, but a benchmark gap versus Anthropic’s Claude Fable 5 narrows when adjusted for tool-free tests.
DeepSeek has quietly swapped its deepseek-v4-pro API offering to a new 0813 build, the general availability version of a model that had been operating as a preview since April, according to Decrypt.
In pricing listings, the model version shown for “deepseek-v4-pro” now reads DeepSeek-V4-Pro-0813, while the posted token rates remain the same: about $0.435 per million input tokens and $0.87 per million output tokens, with cached input at $0.003625.
Decrypt reports that DeepSeek’s internal comparison across nine agent benchmarks where both models are scored places Claude Fable 5 ahead by an average of 5.3%, with DeepSeek winning on two benchmarks. After excluding “Humanity’s Last Exam” without tools, where DeepSeek scores 42.7 versus 53.3 for Fable 5, the remaining rows average a 2.8% gap.
The biggest difference highlighted is cost. Claude Fable 5 is listed at $10 per million input tokens and $50 per million output, while V4 Pro’s blended rates amount to roughly $0.65 versus $30 for Fable 5, about 46 times cheaper, according to the comparison described by Decrypt.