DeepSeek V4.1 Flash: 1M context, MIT weights, ~1/30 Opus cost via Ollama Cloud
Post outlines V4.1-Flash-1017 specs (1M context, MIT weights, $0.15/$0.60 per 1M tokens) and private access via Ollama Cloud, claiming 20–30× cheaper coding-agent runs.
Slash coding-agent costs 20-30× with DeepSeek’s newest Flash model. Released October 17, DeepSeek-V4.1-Flash-1017 is stronger on agentic and coding tasks, has a 1M context, MIT open weights, and costs just $0.15 input / $0.60 output per million tokens off-peak (roughly 1/30th of Claude Opus). It finishes tasks faster and produces shorter outputs than the prior version. Access it privately via Ollama Cloud (US hosting + zero data retention): ollama run deepseek-v4.1-flash:1017-cloud or switch Claude Code / OpenCode instantly with ollama launch claude --model deepseek-v4.1-flash:1017-cloud. Route everyday agents and coding loops here; keep frontier models only for the hardest reasoning. Verified savings on real workloads are large, with no hardware required and strong privacy posture.