AI Briefing#context-window
The latest AI and developer news about #context-window, with the original source and a short summary.
The latest AI and developer news about #context-window, with the original source and a short summary.
Competing statements in context cause 9-110x higher loss than neutral filler, while length alone explains only 2-8% of variance.
The increase from 272k to 500k tokens is available at no extra charge across Codex and ChatGPT Work.
While an LLM context window represents the total token capacity for a single request, the 'Lost in the Middle' phenomenon causes retrieval rates for middle information to drop with longer inputs, making selective management crucial.
Engrim, a local-first SQLite memory engine that preserves context during model switches and session resets in AI CLI environments, has been released.
A KakaoPay developer summarized essential concepts for adopting AI Agents, including the difference between LLMs and Agents, the ReAct pattern, tokens, and context windows.
Research reveals that long contexts alter model activation states, weakening RLHF-based safety constraints.
OpenAI backported a change reducing the model context size from 372k to 272k for the Codex 0.144 release.
OpenAI has lowered the context window size of the Codex model from the existing 372k to 272k.
This is a dictionary summarizing key concepts and terminology related to AI coding, created by Matt Pocock.
llama.cpp has fixed a bug where excessive context checkpoint creation during agent tasks degraded performance.
This introduces a technique that uses a small LLM to remove 68% of RAG context while maintaining 96% recall.
GLM 5.2, now with a context window expanded to 1M tokens, has been added to AI Gateway.
This explains the gap between the advertised context window size of LLMs and the actual effective working range, along with efficient workflows.
Zhipu has released GLM-5.2, a coding-specialized model supporting a 1M context window, and plans to make it open source under an MIT license.
InfiniteKV has been released, dramatically extending a model's context length by compressing the KV cache and storing it in RAM or on disk.
MiniMax has launched the M3 model featuring a 1 million-token context and multimodal capabilities, with weights to be released within 10 days.
It points out MCP's high context cost and operational overhead, suggesting a CLI-first strategy.
A technical analysis has raised concerns that MCP excessively occupies context windows and has low operational reliability.
Anthropic has expanded Claude's context window to 100K tokens, enhancing its ability to analyze large-scale documents.
Providing LLMs with a structural graph of the codebase leads to deeper model exploration, increasing context usage by about 54%.
Anthropic's Claude has been observed encouraging users to sleep, but the exact cause remains unclear.
A new evaluation methodology has been released addressing the 'source-boundary' problem, where LLMs mistake information within context.
A tool has been released that removes duplicate data within LLM context to increase effective token capacity.
Subquadratic touted a 12 million token context window with SSA and an 83 score on MRCR v2, surpassing GPT-5.5.