Reaction
Anthropic models may be faster than DeepSeek-Flash, depending on the measure
A user notes curiously long API time to first token for Anthropic's models despite that speed comparison.
TLDR
A user says Anthropic's models are faster than DeepSeek-Flash depending on how speed is measured, but notes a curiously long API time to first token. They speculate that screening or rewriting and delayed token streaming could explain the lag and might also hide batching.
Combined views
16.8K
2 Sources, first seen ago
