Claude Code context limit bypassed with [1m] suffix
Developers can now bypass Claude Code's default 200k token context limit by appending a [1m] suffix to their model ID. This enables full 1-million-token context utilization for high-capacity models like Qwen 3.6-plus via OpenRouter.
Bypassing the hardcoded 200k limit is a game-changer for massive monorepos where context window pressure is the primary bottleneck. The [1m] suffix explicitly instructs the internal orchestrator to skip auto-compaction, allowing sessions to grow to the model's native capacity. This configuration is essential for models like Qwen 3.6-plus, which offer huge context windows but are often throttled by default CLI settings. Developers using OpenRouter can implement this immediately via environment variables or startup flags. Users should be wary of increased latency and API costs, as a 1M token context can significantly slow down response times and consume credits quickly. This "hidden" flag reflects Anthropic's power-user features that are circulating in the developer community before official documentation updates.
DISCOVERED
8d ago
2026-04-03
PUBLISHED
8d ago
2026-04-03
RELEVANCE
AUTHOR
Mobile_Marsupial_619