DeepSeek's Flash Gets Faster — and Gains Eyes and a Search Engine
DeepSeek-V4.1-Flash-1017 arrives with sharper coding benchmarks, a new KV cache architecture, and a pair of Search and Fetch APIs that undercut every major rival — all while the lab navigates back-to-back service outages that reveal the strain of running the world's cheapest frontier inference at scale.
Sophia Chen