kapynAI / Models

Deepseek V4.1 Flash 1017 beats larger rivals in agentic tasks while once again massively cutting per-task costs

Deepseek launches V4.1‑Flash‑1017, an open‑weight model that outperforms larger rivals on agentic tasks while cutting per‑task costs. The new version delivers faster completion, shorter outputs, and better benchmarks, and Deepseek opens its internal search engine via Search and Fetch APIs at $0.75 per 1,000 calls. Claude Code and Codex users receive free access for now.

The Decoder·Oct 17, 2026

Opening Kapyn…