DeepSeek V4 Pro 0813 is a large-scale mixture-of-experts (MoE) model released on August 12, 2026, featuring a substantial context window of 1,048,576 tokens and support for up to 943,718 completion tokens. Available via OpenRouter, the model supports tool calling and structured outputs through JSON schema. Benchmarking data from Artificial Analysis for the "Reasoning, Max Effort" configuration shows high performance in specialized domains, including a 92.8% score on GPQA Diamond, an 80.3% score on AA-LCR, and a 68.8% Coding Index. Other significant metrics include a 41.3% Agentic Index, a 51.0% SciCode score, and a 41.0% HLE score. Economically, the model is priced at $0.2528 per million input tokens and $2.88 per million output tokens, with a reduced rate of $0.08832 per million tokens for cache reads. OpenRouter facilitates access through various routing modes—Balanced, Nitro, Floor, and Exacto—and manages 22 different providers, including DeepSeek, Baidu Qianfan, and NovitaAI. Over a 3-day period, the service demonstrated 100% uptime and 99.78% availability. The model is compatible with OpenAI-style SDKs by updating the model slug and base URL.