Skip to content

OpenCode token usage is reported as zero in evaluation aggregates #421

Description

@ChathurangiShyalika

Description

OpenCode trajectories include token usage in raw events under tokens.input, tokens.output, tokens.reasoning, and tokens.cache, but the current parser only reads SDK-style fields like input_tokens and output_tokens.

Because of this, OpenCode reports show Tokens In = 0 and Tokens Out = 0, making aggregate comparisons with other CLI agents inaccurate.

Expected Behavior

OpenCode token usage should be parsed from raw OpenCode events and included in evaluation metrics.

Acceptance Criteria

  • OpenCode runner records nonzero token counts from OpenCode tokens events.
  • Evaluation metrics can recover token usage from existing OpenCode trajectories.
  • Aggregate reports show correct OpenCode token totals.

Metadata

Metadata

Labels

No labels
No labels

Type

No type

Fields

No fields configured for issues without a type.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions