deepseek-ai/DeepSeek-V4-Pro

Timeline

1. Oct 16, 2026 · Research paper · 1 source
   PresenceBench: An Exploratory Evaluation of Large Language Model Adaptation to User State
   Presence-aware intelligence is the capacity of an AI system's intelligence layer to adapt appropriately to the functional state of the person it serves (here readiness, flow, and stability) while preserving task quality and safety.
2. Oct 10, 2026 · Research paper · 1 source
   Trace Evolution Bench: Measuring Whether Agents Improve from Their Own Environment Traces
   We introduce Trace Evolution Bench (TEB), a protocol that separates that rescore from a context or weight update, together with three frozen 80/20/40 packs.
3. Oct 9, 2026 · Research paper · 1 source
   Not All Correct Fixes Run Alike: An Empirical Study of Patch Performance in Repository-Level Issue Resolution
   Repository-level issue-resolution benchmarks such as SWE-bench evaluate submitted patches primarily based on whether they pass the correctness tests, leaving the runtime performance of different test-passing patches largely unexplored.
4. Oct 2, 2026 · Opinion / analysis · 1 source
   Hundreds of millions of AI agents are coming. Is there work for them?
   AI companies are shelling out hundreds of billions of dollars each year on chips and data centers, betting that the massive compute buildouts will run AI agents capable of doing work that people do today.

Often appears with