Long context is Thinking Machines' biggest flaw, dev says in follow-up
gnukeith · x · 2026-11-22
In a follow-up, gnukeith pushes back on 'it runs on a 5090' replies: scaling far beyond a 512 context window matters, and the model's RAM footprint forces constant compaction instead of running multiple agents or another model. He says it takes as much memory as a Qwen 3.8 27B and calls long-context efficiency Thinking Machines' biggest flaw.
Related event: Developers Criticize Inkling 2 mini's Memory-Hungry Long Context(2 posts)→
More from coding & agent
- Hacker tricked Meta's Muse agent into silently stealing encrypted passwords — AccBalanced · 2026-11-22
- Dev uses Claude with his game engine plus fal, Tripo, Meshy and ElevenLabs to make a movie trailer — gorkem · 2026-11-22
- MIT study: coding agents boost commits 240% but releases only 30% — apps ship into the void — GaryMarcus · 2026-11-22
- Agent failures come from broken loops, not bad reasoning: the check step is the real job — AgentSearchEngine · 2026-11-22
- gemini-cli PR fixes ACP mode hanging on initialize when sandbox is enabled — Raamia · 2026-11-22
- Open-source coding agent control plane T3 Code claims 500K+ devs orchestrating Claude Code, Codex and more — churchkey · 2026-11-22