Why CEREBRO kept it

Claims 40% usage savings in coding agent; no mechanism shown.

Backlinks

Appeared in 1 briefing

• 2026-11-07

Related

Shares tags: ai/llm-mechanics · cerebro/signal

• RT @AdamHoltererer: Opus 5.5 fails twice as many T3 turns on max as on high My first guess would be that it's because turns on max, take longer, but no: max fails more in every turn-length bucket, anx · 2026-11-07
• I gave Adam access to the T3 Code anonymized telemetry data to make more posts like this :) Anyways, more reasons to never use 'max' reasoning:x · 2026-11-07
• A Jev that can ask Opus for help gives Opus accuracy at 4x lower costhackernews · 2026-11-07
• Stanford researchers develop a new AI memory system that recalls information from entire conversation histories using just 16 memory tokens. It reduces document-token usage by up to 4,212×, runs 27× freddit · 2026-11-07
• Qwen3.8-27B Benchmark: 6 Inference Engines, GGUF vs NInfer, 256K Context — Comparing Both Speed AND Qualityreddit · 2026-11-06
• 'You Don't Need Long Context' NVIDIA finds that LLM agents can complete complex long-horizon tasks without maintaining their entire conversation history, which reduces inference costs. So instead ofx · 2026-11-06

Also from x.com

• RT @AdamHoltererer: Opus 5.5 fails twice as many T3 turns on max as on high My first guess would be that it's because turns on max, take longer, but no: max fails more in every turn-length bucket, anx · 2026-11-07
• I gave Adam access to the T3 Code anonymized telemetry data to make more posts like this :) Anyways, more reasons to never use 'max' reasoning:x · 2026-11-07
• Introducing Claude Code for DaVinci Resolve. one prompt. Motion edits the entire video. then it opens in Resolve, every layer a real Fusion node. Resolve editors still cutting by hand are about to gx · 2026-11-07
• Prompt of the day is a simple one: 'How do we make this change fully verifiable?' Makes the agent bend over backwards to verify all its changes. Signing up for real services, clicking through the frx · 2026-11-06
• 'You Don't Need Long Context' NVIDIA finds that LLM agents can complete complex long-horizon tasks without maintaining their entire conversation history, which reduces inference costs. So instead ofx · 2026-11-06
• semianalysis just dropped GDPx: the best AI agent finishes 12% of real expert work the bigger story is #2. gemini 4 carbon lands 1.1 points behind opus 5.5 at about a third of the cost per task scorx · 2026-11-05