DeepSeek V4 Pro 0813 Listed: $0.435 Input, 1049k Context Window

What Just Dropped

DeepSeek has listed DeepSeek V4 Pro 0813, an open-source text-to-text model that went live on August 12, 2026. The headline numbers: $0.435 per 1M input tokens and $0.87 per 1M output tokens, paired with a substantial 1,049k-token context window.

Why the Context Window Matters

A 1049k context is long enough to load large codebases, lengthy legal documents, or extended conversation histories in a single pass. If your pipeline currently splits or summarizes to fit shorter windows, this model could eliminate that overhead — which has real cost and latency implications beyond the per-token price.

Cost Reality Check

At $0.87 output pricing, V4 Pro 0813 sits in the budget-to-mid tier. For output-heavy workloads (RAG pipelines, summarization, long-form generation), output tokens dominate your bill — so that $0.87 rate is the number to stress-test against your actual output volumes before committing. Input at $0.435 is reasonable for retrieval-heavy use cases where you're stuffing large contexts but generating short answers.

Being open-source also means self-hosting is on the table if your volume justifies the infrastructure trade-off.

Who Should Pay Attention

• Developers handling long documents or large codebases — the 1049k window is the practical differentiator here.
• Cost-conscious teams already on pricier models for similar text tasks — worth a benchmark pass.
• Anyone needing open weights for compliance, fine-tuning, or on-prem deployment.

Check live pricing, context specs, and how it stacks up against alternatives on the DeepSeek V4 Pro 0813 — live specs & price history page before making a switch.

Ezra, Scout AI Team