DeepSeek V4 Flash quietly extends 1M context to standard tier — Apache-2.0 weights match closed-flagship reasoning on Pass@1

DeepSeek extended the 1M context window to its V4 Flash tier this week, pushing the cheaper standard SKU into a capability bracket previously occupied only by V4 Pro and closed flagships. Combined with the unchanged 80.6% SWE-Bench Verified ceiling and the MIT/Apache-2.0 license, the practical effect is to compress the price-quality gradient on long-context production workloads.

The benchmark headline is misleading at first read; capability gains on Pass@k are small. The procurement-relevant gain is on Pass@1 at long contexts — the only metric that actually matters for the production-inference cost equation. DeepSeek V4 Flash now answers correctly on the first attempt across a substantially wider range of long-document workloads than the prior generation, at the same token price.

The aggregation effect compounds the trend we tracked earlier — Chinese open-weight share of OpenRouter usage now exceeds 60%, and Flash tiers eat further into the closed-flagship justification. The Q3 2026 question is whether DeepSeek pushes Pro-tier pricing down too, or holds Pro as the premium SKU and lets Flash absorb most production traffic.

HuggingFace — DeepSeek V4 release notes → · Codersera — best open-source LLM May 2026 →