DeepSeek V4 Flash + Pro from April 24 — MIT license, 1M-token context, two-month-old release still defines current open-source frontier alongside GLM-5.2 and MiniMax M3
DeepSeek V4 Flash and V4 Pro (released April 24 2026) remain current open-source frontier offerings two months after release — MIT license, 1M-token context across both tiers. The sustained relevance against newer releases (GLM-5.2 June 13, MiniMax M3 June 1, Qwen 3.5) reflects how durable DeepSeek's April release was at the time. The two-month sustained-leadership matters for procurement-cadence planning.
The substantive piece is the sustained-leadership pattern in a high-release-cadence environment. Open-source releases through H1 2026 averaged one major release every 2-3 days. DeepSeek V4's April 24 release remained at the frontier through GLM-5.2 (June 13), MiniMax M3 (June 1), Qwen 3.5 — a 2-month sustained-relevance window that's substantively durable for the open-weight category. The durability matters for procurement-cadence planning.
The competitive read for H2 2026 open-source procurement is that the dual-tier MoE architecture (V4-Pro 1.6T total / V4-Flash 284B total) provides workload-shape flexibility within a single vendor relationship. MiniMax M3's multi-dimension capability claim and GLM-5.2's Intelligence Index leadership represent newer entrants competing against DeepSeek V4's established procurement position.
LLM Stats — AI Updates Today (June 2026) → · Kilo AI — Best Open-Source & Open-Weight Coding Models (2026) →