Million-Token Context
Million-Token Context Is Practical, but Retrieval Engineering Still Matters
Million-token windows enable cross-file comparison, but reliable results still depend on retrieval-first workflows, long-context verification, and key
Back to home
Long Context
4 articles tagged with this topic
Ling-3.1
Ling-3.1 Flash Goes Open Source After Two-Week Free Run: 560B Total, 25B Active
Chinese team releases Ling-3.1-flash: 560B total params, 25B activated per call, 1M token context. Free for two weeks, then open source.
Qwen
Apple M5 Max hits 350K-token context — hardware bar cleared, quality still lags
128GB MacBook Pro holds Qwen3.8 at 350K tokens, but past 100K the model 'confuses roles.' Hardware bar falls; quality still lags.
Qwen3
Qwen 27B Long-Context Speed Drops 65% — Local Players Distrust Benchmarks
AMD's 51.8 tokens/sec Qwen3 27B claim is ideal-only, Reddit user finds; real long-context drops 75→26 (–65%). Local LLMs now judged on stability.