Alan-Jowett / CoPilot-For-Consensus / 20758856212 / 26
79%
main: 79%

DEFAULT BRANCH: main
Ran
Files 6
Run time 0s
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 72.571%. Remained the same
20758856212.26

push

github

web-flow 
fix: Switch to gpt-4o-mini due to gpt-4o quota exhaustion (#772)

* fix: Switch to gpt-4o-mini due to gpt-4o quota exhaustion

- Changed model from gpt-4o to gpt-4o-mini in openai.bicep
- Added 2024-07-18 model version support (required for gpt-4o-mini)
- Updated parameters.dev.json to use 2024-07-18 version
- Changed default region to westus2 in parameters

Rationale:
- westus gpt-4o quota exhausted (50K/50K TPM from phantom allocations)
- gpt-4o-mini has 200K TPM available in westus
- gpt-4o-mini provides 80-90% of gpt-4o capability with 128K context
- Other regions lack OpenAI subscription approval

Signed-off-by: GitHub Copilot <copilot@github.com>

* fix: Add missing openai dependency for Azure OpenAI integration

- Install copilot_embedding[openai] extra in embedding Dockerfile
- Install copilot_summarization[openai] extra in summarization Dockerfile

Fixes startup failures in Azure deployment:
- embedding: 'openai is required for OpenAIEmbeddingProvider'
- summarization: 'openai is required for OpenAISummarizer'

Both services need the openai Python package to use Azure OpenAI.

Signed-off-by: GitHub Copilot <copilot@github.com>

* fix: Add openai dependency to reporting service

Reporting service also uses copilot_embedding and needs Azure OpenAI support.

Signed-off-by: GitHub Copilot <copilot@github.com>

* refactor: Parameterize OpenAI model name for per-environment configuration

- Add azureOpenAIModelName parameter (gpt-4o or gpt-4o-mini)
- Revert default model version to 2024-11-20 (for gpt-4o)
- Dev environment explicitly uses gpt-4o-mini with 2024-07-18
- Staging/prod can continue using gpt-4o with 2024-11-20

This addresses PR review feedback: staging/prod parameter files
specify 2024-11-20 (gpt-4o version) but the model was hardcoded
to gpt-4o-mini. Parameterizing the model name allows each
environment to choose the appropriate model and version combination.

Closes review comments on PR #772.

Signed-off-by: GitHub Copilot <copilot@... (continued)

635 of 875 relevant lines covered (72.57%)

0.73 hits per line

Source Files on job copilot_storage - 20758856212.26