DeepSeek V4.1 Flash 1017: post-trained update with benchmark and API pricing details
Notes that 1017 is a post-trained update, not a new architecture: HLE 42.6 (from 36.8), Terminal-Bench 42.3 (from 31.2), DeepSWE 77.0 (from 74.2), same deepseek-flash API at $0.003 cache hit, $0.15 miss, $0.60 output.
DeepSeek-V4.1-Flash-1017 is out — not a new architecture, post-trained. Stronger benches, faster tasks, shorter outputs. Live on web / app / API; same calls and pricing. Official vs last Flash: HLE 42.6 (was 36.8), Terminal-Bench 4.0 → 42.3 (was 31.2), DeepSWE 77.0 (was 74.2), etc. API still `deepseek-flash` (now pointing at 1017). OpenRouter snapshot: `deepseek/deepseek-v4.1-flash-1017` (~$0.15 / $0.60 per M tokens; cache reads cheaper). Official idle: cache hit $0.003 / miss $0.15 / out $0.60; peak ~2×.
This post quotes another post: x.com/deepseek_ai/status/2111375640717033777