cost subcommand

[Cedric / ViaDézo1er (viadezo1er)]

STACKED BARS for bt cost --group-by user --group-by hour

say the cost is only for tokens not for storage cost

create a skill to teach the agent to use btql to see the data however they want, because we're not able to think of all the ways customers will want to use our data

TODO

repeated --group-by
too few abscissa legends in images
legends too small in images
don't separate the cost depending on the source unless a flag is passed, otherwise bt cost --group-by user shows the same user multiple times. maybe only sort depending on active/passive (eg coding agent vs topics)?
i think there is a massive undercounting, i almost don't appear

caching?

Rebased on cedric/prune-profiles to dogfood.
#274 will be merged before this pr.

bt cost logs:

7 cost groups in BT Staging / cedric-traces
2026-07-14T01:35:49Z to 2026-07-21T01:35:49Z

Model                        Purpose    Cost       Coverage     Sources                          
claude-opus-4-8              default    $48.40     430/430      Braintrust 430                   
zai-org/GLM-5.2              default    $45.49     2267/2267    Braintrust 2267, no usage 272    
gpt-5.5                      default    $32.57     705/705      Braintrust 705, no usage 9       
gpt-5.6-sol                  default    $14.97     741/741      Braintrust 741, no usage 11      
claude-haiku-4-5-20251001    default    $0.339     50/50        Braintrust 50                    
<missing>                    default    unknown    0/4          —                                
glm-5.2                      default    unknown    0/1749       no usage 11                      

Total: ~$141.76  1753 unpriced token spans  303 without cost or usage

bt cost logs --json|jq:

{
  "project": "cedric-traces",
  "org": "BT Staging",
  "currency": "USD",
  "since": "2026-07-14T01:35:29Z",
  "until": "2026-07-21T01:35:29Z",
  "excludes_scorers": false,
  "pricing_file": null,
  "rows": [
    {
      "model": "claude-opus-4-8",
      "purpose": null,
      "candidate_spans": 430,
      "cost": 48.40370225,
      "braintrust_cost": 48.40370225,
      "file_cost": null,
      "braintrust_priced_spans": 430,
      "file_priced_spans": 0,
      "unpriced_token_spans": 0,
      "no_usage_spans": 0,
      "coverage": "full"
    },
    {
      "model": "zai-org/GLM-5.2",
      "purpose": null,
      "candidate_spans": 2539,
      "cost": 45.48671468,
      "braintrust_cost": 45.48671468,
      "file_cost": null,
      "braintrust_priced_spans": 2267,
      "file_priced_spans": 0,
      "unpriced_token_spans": 0,
      "no_usage_spans": 272,
      "coverage": "full"
    },
    {
      "model": "gpt-5.5",
      "purpose": null,
      "candidate_spans": 714,
      "cost": 32.5653,
      "braintrust_cost": 32.5653,
      "file_cost": null,
      "braintrust_priced_spans": 705,
      "file_priced_spans": 0,
      "unpriced_token_spans": 0,
      "no_usage_spans": 9,
      "coverage": "full"
    },
    {
      "model": "gpt-5.6-sol",
      "purpose": null,
      "candidate_spans": 752,
      "cost": 14.96779075,
      "braintrust_cost": 14.96779075,
      "file_cost": null,
      "braintrust_priced_spans": 741,
      "file_priced_spans": 0,
      "unpriced_token_spans": 0,
      "no_usage_spans": 11,
      "coverage": "full"
    },
    {
      "model": "claude-haiku-4-5-20251001",
      "purpose": null,
      "candidate_spans": 50,
      "cost": 0.33857415,
      "braintrust_cost": 0.33857415,
      "file_cost": null,
      "braintrust_priced_spans": 50,
      "file_priced_spans": 0,
      "unpriced_token_spans": 0,
      "no_usage_spans": 0,
      "coverage": "full"
    },
    {
      "model": null,
      "purpose": null,
      "candidate_spans": 4,
      "cost": null,
      "braintrust_cost": null,
      "file_cost": null,
      "braintrust_priced_spans": 0,
      "file_priced_spans": 0,
      "unpriced_token_spans": 4,
      "no_usage_spans": 0,
      "coverage": "unknown"
    },
    {
      "model": "glm-5.2",
      "purpose": null,
      "candidate_spans": 1760,
      "cost": null,
      "braintrust_cost": null,
      "file_cost": null,
      "braintrust_priced_spans": 0,
      "file_priced_spans": 0,
      "unpriced_token_spans": 1749,
      "no_usage_spans": 11,
      "coverage": "unknown"
    }
  ],
  "totals": {
    "candidate_spans": 6249,
    "cost": 141.76208183,
    "braintrust_cost": 141.76208183,
    "file_cost": null,
    "braintrust_priced_spans": 4193,
    "file_priced_spans": 0,
    "unpriced_token_spans": 1753,
    "no_usage_spans": 303,
    "coverage": "partial"
  }
}

[github-actions]

Latest downloadable build artifacts for this PR commit 0c407076eb84:

• Workflow run: https://github.com/braintrustdata/bt/actions/runs/29793424548
• Download all artifacts (GitHub CLI): gh run download 29793424548 --repo braintrustdata/bt
• Installers are published from main automatically. To publish one for a PR branch, run release-canary manually via workflow_dispatch.

Available artifact names

• artifacts-build-global
• artifacts-build-local-x86_64-apple-darwin
• artifacts-build-local-aarch64-pc-windows-msvc
• artifacts-build-local-x86_64-pc-windows-msvc
• artifacts-build-local-x86_64-unknown-linux-musl
• artifacts-build-local-x86_64-unknown-linux-gnu
• artifacts-build-local-aarch64-apple-darwin
• artifacts-build-local-aarch64-unknown-linux-gnu
• artifacts-plan-dist-manifest
• cargo-dist-cache