fix(llmobs): fix missing estimated cost on Bedrock LLM spans [MLOB-7008] by heyitsgrace996 · Pull Request #17293 · DataDog/dd-trace-py

heyitsgrace996 · 2026-04-03T00:02:23Z

Problem

Estimated cost is missing or incorrect on Bedrock LLM spans. All Bedrock models tested were impacted - either with no cost data match, or matched direct-API pricing instead of Bedrock pricing.

Root Cause

parse_model_id() splits model IDs like "amazon.nova-lite-v1:0" into model_provider="amazon" and model_name="nova-lite-v1:0" for APM span tags. The LLMObs integration passed these split values through to the backend cost estimator, which couldn't match either:

Provider "amazon" doesn't match the AWS Bedrock provider (provider_match: contains "bedrock")
Model name "nova-lite-v1:0" doesn't match pricing entries (match: contains "amazon.nova-lite")

For vendors with their own direct-API provider (anthropic, mistral, deepseek, openai), the split provider name matched the wrong provider, resulting in direct-API pricing instead of Bedrock pricing.

Fix

Pass the original model_id through the execution context to LLMObs instead of the split values. Set model_provider to "amazon_bedrock" for all Bedrock LLM spans.

model_provider: "amazon" -> "amazon_bedrock" (matches backend cost estimator and existing UI icon mapping)
model_name: "nova-lite-v1:0" -> "amazon.nova-lite-v1:0" (matches backend pricing entries)

"amazon_bedrock" was chosen over "bedrock" because the frontend already maps "amazon_bedrock" to the LlmType.AMAZON_BEDROCK icon (source). Using "bedrock" would have required a separate frontend change.

APM span tags bedrock.request.model_provider and bedrock.request.model are unchanged.

Testing

Updated assertions in tests/contrib/botocore/test_bedrock_llmobs.py (44 passed, 9 skipped - botocore version gating).

Tested 29 Bedrock models across 7 vendors (amazon, anthropic, meta, mistral, deepseek, openai, qwen) with AWS Bedrock calls. Fix validated locally - confirmed correct cost estimation and UI icon rendering in LLMObs traces explorer.

Codex automated review completed on final commit - no blocking issues found.

Before change:

After change:

Risks

Changes meta.model_provider and meta.model_name on LLMObs span events. Customers filtering by @meta.model_provider:anthropic for Bedrock models would need to update to @meta.model_provider:amazon_bedrock. However, those customers were getting incorrect pricing, so the filter was already misleading.
Internal telemetry model_provider tag changes from vendor names to amazon_bedrock for Bedrock spans, but again before it was not being marked as Bedrock at all.
ARN-format model IDs (e.g., arn:aws:bedrock:us-west-2:...:inference-profile/...) will see the full ARN string as model_name in the LLMObs UI.
Cost matching still works for system inference profiles (the model name is embedded in the ARN and matched via contains clauses).

Release note included.

cit-pr-commenter-54b7da · 2026-04-03T00:03:07Z

Codeowners resolved as

ddtrace/contrib/internal/botocore/services/bedrock.py                   @DataDog/ml-observability
ddtrace/llmobs/_integrations/bedrock.py                                 @DataDog/ml-observability
releasenotes/notes/llmobs-fix-bedrock-cost-estimation-19af7556637be09e.yaml  @DataDog/apm-python
tests/contrib/botocore/test_bedrock_llmobs.py                           @DataDog/ml-observability

heyitsgrace996 · 2026-04-03T00:52:36Z

@codex

chatgpt-codex-connector · 2026-04-03T00:55:58Z

Codex Review: Didn't find any major issues. Keep them coming!

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Pass the original model_id through the execution context to LLMObs instead of the split values from parse_model_id(). Set model_provider to "amazon_bedrock" for all Bedrock spans. All Bedrock models tested were impacted - either with no cost data match, or a direct API provider match resulting in incorrect pricing.

datadog-official · 2026-04-03T01:51:32Z

✅ Tests

🎉 All green!

❄️ No new flaky tests detected
🧪 All tests passed

_{This comment will be updated automatically if new data arrives.

🔗 Commit SHA: 75d46d4 | Docs | Datadog PR Page | Was this helpful? React with 👍/👎 or give us feedback!}

Yun-Kim

Great change! Just a small comment on the release note but otherwise LGTM 👍

…637be09e.yaml

Yun-Kim · 2026-04-03T18:53:26Z

/merge

gh-worker-devflow-routing-ef8351 · 2026-04-03T18:53:31Z

View all feedbacks in Devflow UI.

2026-04-03 18:53:31 UTC ℹ️ Start processing command /merge

2026-04-03 18:53:39 UTC ℹ️ MergeQueue: waiting for PR to be ready

This pull request is not mergeable according to GitHub. Common reasons include pending required checks, missing approvals, or merge conflicts — but it could also be blocked by other repository rules or settings.
It will be added to the queue as soon as checks pass and/or get approvals. View in MergeQueue UI.
Note: if you pushed new commits since the last approval, you may need additional approval.
You can remove it from the waiting list with /remove command.

2026-04-03 19:12:09 UTC ℹ️ MergeQueue: merge request added to the queue

The expected merge time in main is approximately 57m (p90).

2026-04-03 21:12:56 UTC ❌ MergeQueue: The build pipeline has timeout

The merge request has been interrupted because the build 106012690 took longer than expected. The current limit for the base branch 'main' is 120 minutes.

Yun-Kim · 2026-04-03T22:41:53Z

/merge

gh-worker-devflow-routing-ef8351 · 2026-04-03T22:41:58Z

View all feedbacks in Devflow UI.

2026-04-03 22:41:57 UTC ℹ️ Start processing command /merge

2026-04-03 22:42:01 UTC ℹ️ MergeQueue: pull request added to the queue

The expected merge time in main is approximately 57m (p90).

2026-04-03 23:22:18 UTC ℹ️ MergeQueue: This merge request was merged

…08] (#17293) ## Problem Estimated cost is missing or incorrect on Bedrock LLM spans. All Bedrock models tested were impacted - either with no cost data match, or matched direct-API pricing instead of Bedrock pricing. ## Root Cause `parse_model_id()` splits model IDs like `"amazon.nova-lite-v1:0"` into `model_provider="amazon"` and `model_name="nova-lite-v1:0"` for APM span tags. The LLMObs integration passed these split values through to the backend cost estimator, which couldn't match either: - Provider `"amazon"` doesn't match the AWS Bedrock provider (`provider_match: contains "bedrock"`) - Model name `"nova-lite-v1:0"` doesn't match pricing entries (`match: contains "amazon.nova-lite"`) For vendors with their own direct-API provider (anthropic, mistral, deepseek, openai), the split provider name matched the wrong provider, resulting in direct-API pricing instead of Bedrock pricing. ## Fix Pass the original `model_id` through the execution context to LLMObs instead of the split values. Set `model_provider` to `"amazon_bedrock"` for all Bedrock LLM spans. - `model_provider`: `"amazon"` -> `"amazon_bedrock"` (matches backend cost estimator and existing UI icon mapping) - `model_name`: `"nova-lite-v1:0"` -> `"amazon.nova-lite-v1:0"` (matches backend pricing entries) `"amazon_bedrock"` was chosen over `"bedrock"` because the frontend already maps `"amazon_bedrock"` to the `LlmType.AMAZON_BEDROCK` icon ([source](https://github.com/DataDog/web-ui/blob/prod/packages/apps/llm/lib/utils/get-llm-type-from-source.ts)). Using `"bedrock"` would have required a separate frontend change. APM span tags `bedrock.request.model_provider` and `bedrock.request.model` are unchanged. ## Testing Updated assertions in `tests/contrib/botocore/test_bedrock_llmobs.py` (44 passed, 9 skipped - botocore version gating). Tested 29 Bedrock models across 7 vendors (amazon, anthropic, meta, mistral, deepseek, openai, qwen) with AWS Bedrock calls. Fix validated locally - confirmed correct cost estimation and UI icon rendering in LLMObs traces explorer. Codex automated review completed on final commit - no blocking issues found. **Before change:** <img width="1057" height="598" alt="image" src="https://github.com/user-attachments/assets/8dfc3403-bc78-4f8f-b08b-028c8253dba0" /> <img width="1114" height="621" alt="image" src="https://github.com/user-attachments/assets/3893f580-51c6-4849-b091-24f428d81552" /> **After change:** <img width="1058" height="614" alt="image" src="https://github.com/user-attachments/assets/a37c65e8-47f1-41a1-8348-077313adfefa" /> <img width="1116" height="671" alt="image" src="https://github.com/user-attachments/assets/061f75e6-d01a-4b1b-a42f-076bccfee083" /> ## Risks - Changes `meta.model_provider` and `meta.model_name` on LLMObs span events. Customers filtering by `@meta.model_provider:anthropic` for Bedrock models would need to update to `@meta.model_provider:amazon_bedrock`. However, those customers were getting incorrect pricing, so the filter was already misleading. - Internal telemetry `model_provider` tag changes from vendor names to `amazon_bedrock` for Bedrock spans, but again before it was not being marked as Bedrock at all. - ARN-format model IDs (e.g., `arn:aws:bedrock:us-west-2:...:inference-profile/...`) will see the full ARN string as `model_name` in the LLMObs UI. Cost matching still works for system inference profiles (the model name is embedded in the ARN and matched via `contains` clauses). Release note included. Co-authored-by: Yun-Kim <35776586+Yun-Kim@users.noreply.github.com> Co-authored-by: yun.kim <yun.kim@datadoghq.com>

Pass the full modelId and "amazon_bedrock" as model_provider to LLMObs instead of the split values from parseModelId(). The split values prevented the backend cost estimator from matching Bedrock pricing entries. Mirrors the Python tracer fix in DataDog/dd-trace-py#17293.

heyitsgrace996 force-pushed the grace.williams/fix-bedrock-cost-mlob-7008 branch 5 times, most recently from e0d0215 to 928513f Compare April 3, 2026 00:32

heyitsgrace996 marked this pull request as ready for review April 3, 2026 00:38

heyitsgrace996 requested review from a team as code owners April 3, 2026 00:38

heyitsgrace996 requested review from Kyle-Verhoog and duncanista April 3, 2026 00:38

heyitsgrace996 force-pushed the grace.williams/fix-bedrock-cost-mlob-7008 branch from 928513f to 2c51872 Compare April 3, 2026 01:39

heyitsgrace996 requested review from XG-xin and Yun-Kim April 3, 2026 08:06

Yun-Kim approved these changes Apr 3, 2026

View reviewed changes

Comment thread releasenotes/notes/llmobs-fix-bedrock-cost-estimation-19af7556637be09e.yaml Outdated

Update releasenotes/notes/llmobs-fix-bedrock-cost-estimation-19af7556…

75d46d4

…637be09e.yaml

gh-worker-dd-mergequeue-cf854d Bot merged commit 005967b into main Apr 3, 2026
537 of 539 checks passed

gh-worker-dd-mergequeue-cf854d Bot deleted the grace.williams/fix-bedrock-cost-mlob-7008 branch April 3, 2026 23:22

heyitsgrace996 mentioned this pull request Apr 8, 2026

fix(llmobs): fix missing estimated cost on Bedrock LLM spans DataDog/dd-trace-js#7952

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix(llmobs): fix missing estimated cost on Bedrock LLM spans [MLOB-7008]#17293

fix(llmobs): fix missing estimated cost on Bedrock LLM spans [MLOB-7008]#17293
gh-worker-dd-mergequeue-cf854d[bot] merged 2 commits intomainfrom
grace.williams/fix-bedrock-cost-mlob-7008

heyitsgrace996 commented Apr 3, 2026 •

edited

Loading

Uh oh!

cit-pr-commenter-54b7da Bot commented Apr 3, 2026 •

edited

Loading

Uh oh!

heyitsgrace996 commented Apr 3, 2026

Uh oh!

chatgpt-codex-connector Bot commented Apr 3, 2026

Uh oh!

datadog-official Bot commented Apr 3, 2026 •

edited by datadog-datadog-prod-us1-2 Bot

Loading

Uh oh!

Yun-Kim left a comment

Uh oh!

Uh oh!

Yun-Kim commented Apr 3, 2026

Uh oh!

gh-worker-devflow-routing-ef8351 Bot commented Apr 3, 2026 •

edited

Loading

Uh oh!

Yun-Kim commented Apr 3, 2026

Uh oh!

gh-worker-devflow-routing-ef8351 Bot commented Apr 3, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

heyitsgrace996 commented Apr 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Problem

Root Cause

Fix

Testing

Risks

Uh oh!

cit-pr-commenter-54b7da Bot commented Apr 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codeowners resolved as

Uh oh!

heyitsgrace996 commented Apr 3, 2026

Uh oh!

chatgpt-codex-connector Bot commented Apr 3, 2026

Uh oh!

datadog-official Bot commented Apr 3, 2026 • edited by datadog-datadog-prod-us1-2 Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Yun-Kim left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Yun-Kim commented Apr 3, 2026

Uh oh!

gh-worker-devflow-routing-ef8351 Bot commented Apr 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Yun-Kim commented Apr 3, 2026

Uh oh!

gh-worker-devflow-routing-ef8351 Bot commented Apr 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

heyitsgrace996 commented Apr 3, 2026 •

edited

Loading

cit-pr-commenter-54b7da Bot commented Apr 3, 2026 •

edited

Loading

datadog-official Bot commented Apr 3, 2026 •

edited by datadog-datadog-prod-us1-2 Bot

Loading

gh-worker-devflow-routing-ef8351 Bot commented Apr 3, 2026 •

edited

Loading

gh-worker-devflow-routing-ef8351 Bot commented Apr 3, 2026 •

edited

Loading