Release Event And Versioning

Issue 90 Edition 2026-03-31 3 min read

Not accepted General

Sources: 1 • Confidence: Medium • Updated: 2026-04-01 03:37

Key takeaways

llm-echo version 0.4 has been released.
llm-echo prompt responses now include populated input_tokens and output_tokens fields.

llm-echo prompt responses now include populated input_tokens and output_tokens fields.

What are the complete llm-echo 0.4 release notes/changelog items beyond token field population?
How are input_tokens and output_tokens computed (provider-returned vs. estimated), and what is the exact counting convention used?
Across which providers/models does llm-echo 0.4 return non-null token fields, and are there known exceptions?
Are there any migration steps or compatibility constraints when upgrading from llm-echo 0.3 to 0.4?
Is there any direct operator/product/investor decision-readthrough described for these deltas (e.g., changes to billing, rate limiting, observability tooling, or SLAs)?

Populated input_tokens and output_tokens could enable clearer usage metering and cost analytics, improving observability and internal chargeback workflows for users who rely on token accounting.
Version 0.4 release may signal a broader maturation step beyond token fields, potentially impacting integration stability, response schemas, or compatibility expectations for downstream tooling.
If token fields are consistent across providers and models, llm-echo could become more attractive for multi-provider monitoring and governance use cases that need standardized usage metadata.

Full 0.4 changelog shows additional production-focused improvements such as schema stability guarantees, compatibility notes, or operational features beyond token field population.
Documentation specifies token counting conventions and whether counts are provider-returned or estimated, plus coverage across providers and models with minimal exceptions.
User or operator guidance highlights direct impacts such as improved billing reconciliation, rate limiting, analytics dashboards, or SLA reporting enabled by token metadata.

Token fields are frequently null or inconsistent across major providers and models, or rely on unreliable estimation, limiting usefulness for metering and analytics.
Upgrade from 0.3 to 0.4 introduces breaking changes or migration burden that reduces adoption of the new version despite the added metadata.
Release notes indicate the change is narrowly cosmetic with no reliable operational read-through, and users do not adopt token fields in workflows.