Tokens Boutique

A batch of changes went in today. Here's what's new.

FEATThe output token field in the calculator is now capped at 384k and the model table can be sorted by max output.

FEATSorting also defaults to most capable first, then cheapest.

FEATAll prices shown are cache-miss prices. We plan to add cache-hit pricing next.

FEATAdded Perplexity as a vendor, along with Claude Fable 5 and Claude Sonnet 5.

FIXClaude Sonnet 4.6's max output was listed as 64K, it's actually 128K.

FIXGPT-5.6 Sol, GPT-5.6 Terra, GPT-5.6 Luna and GPT-5.4 had the wrong context window listed.

FIXDeepSeek V3.2-Exp, DeepSeek V3.1 and DeepSeek R1 had incorrect context or max output values.

FEATOn mobile, the Discord button and the price trend chart are now hidden to keep things clean.

FEATAnd, well, you're reading this on the new blog!