test(ci): refresh retired OpenAI tool-call models

[yuneng-berri]

TLDR

Problem this solves:

• Four live tool-call cases use a retired OpenAI model

How it solves it:

• Select a supported model and retain live provider calls
• Assert returned tool calls are present and correctly named
• Preserve sampling parameters with explicit non-reasoning mode

This PR is now batch 1 only, limited to three test files. Legacy completion fakes, callback resets, schema snapshots, warning filters, and other fixture repairs have been removed from its diff

The streaming follow-up retains temperature=0.2 and seed=22, with reasoning_effort="none". A fresh real-provider call confirmed that combination is accepted. No live call is replaced with a fake

Validation

All four affected live cases pass with retries disabled. make check passes the test-tree lint and test-quality budget gates. The replacement model exercises the Responses-to-chat transformation. Dropping returned tool calls in that active response transformation makes both added assertions fail; restoring the product passes all four cases. Hosted checks and reviews must qualify the new head independently

The two logging-file cases assert returned tool calls. These assertions do not establish delivery to the external logging services

Pre-Submission checklist

• I have added meaningful tests
• The handful of test files covering my change pass locally
• My PR passes all required CI/CD checks
• My PR's scope is as isolated as possible
• I have received the required current-head Greptile confidence score

Screenshots / Proof of Fix

Use the isolated proxy at http://127.0.0.1:49548. Register ci-tool-proof through POST /model/new, setting litellm_params.model to the model under test and supplying the provider key privately. Both cases use the same weather-tool request below. Delete the temporary deployment afterward

curl -sS http://127.0.0.1:49548/v1/chat/completions \
  -H "Authorization: Bearer $LITELLM_MASTER_KEY" \
  -H 'Content-Type: application/json' \
  --data-binary @- <<'JSON'
{"model":"ci-tool-proof","messages":[{"role":"user","content":"What's the weather like in San Francisco, Tokyo, and Paris?"}],"tools":[{"type":"function","function":{"name":"get_current_weather","description":"Get the current weather in a given location","parameters":{"type":"object","properties":{"location":{"type":"string"},"unit":{"type":"string","enum":["celsius","fahrenheit"]}},"required":["location"]}}}],"tool_choice":"auto"}
JSON

Before (9525452)

1. Register openai/gpt-3.5-turbo-1106 and send the shared request
2. Observe HTTP 404 and the provider message that the model is deprecated

After (25b8ced)

1. Register openai/gpt-6-luna and send the shared request
2. Observe HTTP 200, three get_current_weather calls, and 152 total tokens

Caveats

Medium

• CircleCI pipeline creation is currently disabled for this project
• OSV reports advisories in unchanged dependency lockfiles
• The unchanged scheduler stagger test fails in hosted CI
• The miscellaneous unit-test job also fails in hosted CI

Type

Test

Final Attestation

• The tests check the right things, including edge cases, and regressions in the respective real-world customer use-cases are not possible after this PR

[greptile-apps]

Confidence Score: 5/5

[Low risk] Test files update model names and add assertions.

The current three-file PR appears safe to merge based on this review.

Summary

The PR replaces a retired OpenAI model in four live tool-calling cases, adds tool-call assertions to two tests, and preserves sampling parameters in the streaming follow-up. All previous Greptile threads are resolved and concern changes outside the current three-file PR.

_{Reviews (10) · Last reviewed commit: "test(ci): refresh retired OpenAI tool-ca..."}

[yuneng-berri]

@greptileai Please review the latest commit, including the Bedrock routing assertion and its strengthened negative control

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[yuneng-berri]

@greptileai Please review the current head. Both findings are addressed, with request validation and the original credential behavior preserved

[yuneng-berri]

@greptileai Please review current head, including the verified worker-readiness repair and disambiguated response-format contract with retained negative controls

[yuneng-berri]

bugbot run verbose=true

Please review the current head for regressions in test isolation, request validation, and retained behavioral coverage

[cursor]

Stale Bugbot comment from a previous run.

[yuneng-berri]

@greptileai Please review the current head, including the worker response lifecycle and retained failure checks for missing spend records

[yuneng-berri]

bugbot run verbose=true

Please review the current head, including the worker response lifecycle and retained checks for missing spend records

[cursor]

Stale Bugbot comment from a previous run.

[yuneng-berri]

@greptileai Please review the current head, including held requests on both workers and the preserved exact-once survivor logging assertion

[yuneng-berri]

bugbot run verbose=true

Please review the current head, including held requests on both workers and the preserved exact-once survivor logging assertion

[cursor]

Stale Bugbot comment from a previous run.

[yuneng-berri]

@greptileai Please review the current head, including callback test isolation, the fixture package marker, and budget logger scoping with retained exception checks

[yuneng-berri]

bugbot run verbose=true

Please review the current head, including callback test isolation, the fixture package marker, and budget logger scoping with retained exception checks

[cursor]

Stale Bugbot comment from a previous run.

[yuneng-berri]

@greptileai Please review the current head, including resource ownership, nullable callback metadata, and preserved failure assertions in these test-only repairs

[yuneng-berri]

bugbot run verbose=true

Please review the current head for weakened assertions, resource cleanup, and accidental product changes in these test-only repairs

[cursor]

Stale Bugbot comment from a previous run.

[yuneng-berri]

@greptileai Please review the current head, including the budget permission fixture's supported decrease and unchanged authorization expectations across all roles

[yuneng-berri]

bugbot run verbose=true

Please review the current head, especially whether the budget fixture preserves authorization coverage without depending on case execution order

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 4d002f1. Configure here.}

[yuneng-berri]

@greptileai Please review the current head after splitting this PR down to the live model updates