DeepSeek V4 Pro vs Flash — the procurement decision tree clarifies at MIT-licensed weights

DeepSeek's V4 release (April 24) shipped two SKUs: V4-Pro (1.6T total / 49B active parameters, 80.6 SWE-Bench Verified, 90.1 GPQA Diamond) and V4-Flash (284B total / 13B active, 1M context). Both run under the MIT license, both ship at 1M context, and both clear the bar for production deployment on coding and reasoning workloads. The Pro/Flash bifurcation now mirrors the closed-flagship pricing curve at a fraction of the cost.

The procurement-relevant detail is that V4-Flash retained 1M context — a capability previously reserved for V4-Pro. For long-document RAG and codebase-spanning agentic workflows, that erases the structural case for routing to Pro. The Flash-tier becomes the default; Pro stays on the menu for the residual workloads that need the bigger activation budget.

The geographic dimension matters. DeepSeek's Apache/MIT licensing combined with sub-$0.20/M token pricing puts the model in procurement bracket directly competitive with the closed flagships from US labs. Chinese open-weight share now exceeds 60% of OpenRouter usage; V4 Pro/Flash continues the trend rather than initiating it.

HuggingFace — best open-source LLMs May 2026 → · Codersera — open-source LLM landscape 2026 → · Presenc AI — open-source LLM landscape 2026 →