Capability Split: Cost/Allowance Tracking Moved To Separate Plugin

Issue 91 Edition 2026-04-01 4 min read

General

Sources: 1 • Confidence: High • Updated: 2026-04-02 03:46

Key takeaways

datasette-llm-usage 0.2a0 removed features related to allowances and estimated pricing.
datasette-llm-usage now depends on datasette-llm for model configuration.
If the datasette-llm-usage.log_prompts plugin setting is enabled, full prompts, responses, and tool calls can be logged to the internal llm_usage_prompt_log table.
The /-/llm-usage-simple-prompt page was redesigned and now requires the llm-usage-simple-prompt permission to access.
Allowances and estimated pricing functionality is now handled by datasette-llm-accountant instead of datasette-llm-usage.

datasette-llm-usage 0.2a0 removed features related to allowances and estimated pricing.
Allowances and estimated pricing functionality is now handled by datasette-llm-accountant instead of datasette-llm-usage.

If the datasette-llm-usage.log_prompts plugin setting is enabled, full prompts, responses, and tool calls can be logged to the internal llm_usage_prompt_log table.

The /-/llm-usage-simple-prompt page was redesigned and now requires the llm-usage-simple-prompt permission to access.

What version boundary (previous release number and exact prior behavior) defines the before/after for the removed allowances and estimated pricing features?
Does datasette-llm-accountant fully replace the removed allowances/estimated pricing features, and is its data model compatible with existing datasette-llm-usage data?
What exact configuration changes are required in deployments due to the new dependency on datasette-llm for model configuration?
What schema, retention expectations, and security considerations apply to the llm_usage_prompt_log table when prompt logging is enabled?
What roles or permission configuration patterns are recommended to manage access to /-/llm-usage-simple-prompt after the new permission requirement?

Plugin ecosystem is modularizing LLM usage, cost tracking, and accounting into separate components, suggesting a shift toward composable deployments and potentially more third party extensions if adoption continues.
Operational focus is increasing on governance features like permissioned UI access and optional full prompt logging, implying stronger enterprise compliance and auditing needs within this toolchain.
Centralizing model configuration in a shared dependency suggests consolidation of configuration standards across plugins, which could reduce fragmentation and accelerate coordinated upgrades across related packages.

datasette-llm-accountant documentation and releases show feature parity with removed allowances and estimated pricing, plus clear migration guidance from prior datasette-llm-usage behavior.
Adoption signals such as growing install base, active issues, and community guidance around new permissions and prompt logging, indicating the security and auditing changes are being used rather than avoided.
Clear, stable configuration patterns emerge around datasette-llm model configuration, with fewer breaking changes and fewer support questions about configuration migration.

Users report the split causes fragmentation, complex setup, or incompatible data models between usage tracking and accounting, leading to stalled upgrades or abandonment.
Prompt logging to llm_usage_prompt_log creates security or retention concerns without clear controls, leading organizations to disable features or avoid deployment.
The new datasette-llm dependency introduces frequent breaking changes or unclear configuration migration, increasing operational burden and reducing adoption.