Version 2.17.0

What's new

• Choose models with cost in mind — The model picker now shows subtle cost tiers per model. You can see before sending whether a model is relatively low-cost or expensive.
• Use Claude chats more efficiently — Claude models handle recurring context more consistently. This helps especially in longer conversations with a lot of shared context.

For administrators

• Analyse usage costs more precisely — The statistics page shows costs by user and team, with date filters, model breakdowns, platform/API source filters, and CSV export.
• Check model costs concretely — Model settings now show credits per 1M input and output tokens. This makes cost comparisons easier before enabling models.