Codex with litellm vertex AI endpoint serving gpt-oss-120b model produces no output.

What version of Codex is running?

0.46.0

Which model were you using?

gpt-oss-120b

What platform is your computer?

Linux 6.12.38+deb13-amd64 x86_64 unknown

What steps can reproduce the bug?

litellm with gcloud cli and minimal config.yaml as:

1. Setup gcloud cli (We need this ~/.config/gcloud/application_default_credentials.json). Also enable the Vertex AI API and the gpt oss maas api on Vertex AI.

1. Configure litellm with vertex endpoint:

services:
  litellm:
    image: ghcr.io/berriai/litellm:main-stable
    volumes:
      - ./config.yaml:/app/config.yaml
      - ~/.config/gcloud/application_default_credentials.json:/app/application_default_credentials.json
    command:
      - "--config=/app/config.yaml"
    ports:
      - "4000:4000" # Map the container port to the host, change the host port if necessary
    environment:
      DATABASE_URL: "postgresql://llmproxy:password@db:5432/litellm"
      STORE_MODEL_IN_DB: "True" # allows adding models to proxy via UI
    env_file:
      - .env # Load local .env file
    depends_on:
      - db  # Indicates that this service depends on the 'db' service, ensuring 'db' starts first
    healthcheck:  # Defines the health check configuration for the container
      test: [ "CMD-SHELL", "wget --no-verbose --tries=1 http://localhost:4000/health/liveliness || exit 1" ]  # Command to execute for health check
      interval: 30s  # Perform health check every 30 seconds
      timeout: 10s   # Health check command times out after 10 seconds
      retries: 3     # Retry up to 3 times if health check fails
      start_period: 40s  # Wait 40 seconds after container start before beginning health checks

  db:
    image: postgres:16
    restart: always
    container_name: litellm_db
    environment:
      POSTGRES_DB: litellm
      POSTGRES_USER: llmproxy
      POSTGRES_PASSWORD: password
    ports:
      - "5432:5432"
    volumes:
      - postgres_data:/var/lib/postgresql/data # Persists Postgres data across container restarts
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -d litellm -U llmproxy"]
      interval: 1s
      timeout: 5s
      retries: 10

  prometheus:
    image: prom/prometheus
    volumes:
      - prometheus_data:/prometheus
      - ./prometheus.yml:/etc/prometheus/prometheus.yml
    ports:
      - "9090:9090"
    command:
      - "--config.file=/etc/prometheus/prometheus.yml"
      - "--storage.tsdb.path=/prometheus"
      - "--storage.tsdb.retention.time=15d"
    restart: always

volumes:
  prometheus_data:
    driver: local
  postgres_data:
    name: litellm_postgres_data # Named volume for Postgres data persistence

1. Before running docker compose up, create the required config.yaml (required by our docker compose file) pointing to the models:

model_list:
    - model_name: gpt-oss-120b-litellm
      litellm_params:
        model: vertex_ai/openai/gpt-oss-120b-maas
        vertex_ai_project: "PROJECT_ID"
        vertex_credentials: "/app/application_default_credentials.json"
        #use_in_pass_through: true
      model_info:
        supports_reasoning: True
    - model_name: gpt-oss-20b-litellm
      litellm_params:
        model: vertex_ai/openai/gpt-oss-20b-maas
        vertex_ai_project: "PROJECT_ID"
        vertex_credentials: "/app/application_default_credentials.json"
        #use_in_pass_through: true
      model_info:
        supports_reasoning: True

litellm_settings:
  # Necessary to use the models in codex as Vertex AI endpoint does not
  # support the entire OpenAI Spec of tooluse.
  drop_params: true

And a .env file

LITELLM_MASTER_KEY="sk-1234"

1. Launch codex and test

# Point to your LiteLLM Proxy server
export OPENAI_BASE_URL=http://0.0.0.0:4000 

# Use your LiteLLM API key (if you've set up authentication)
export OPENAI_API_KEY="sk-1234"

codex --model gpt-oss-120b-litellm
# ask something
# task completes but no output

What is the expected behavior?

codex --model gpt-oss-120b-litellm
# ask something
# task completes WITH OUTPUT

What do you see instead?

codex --model gpt-oss-120b-litellm
# ask something
# task completes but no output

Additional information

Linking issue with litellm: https://github.com/BerriAI/litellm/issues/15520

It is somewhat involved to setup vertex due to the gcloud cli authentication and enabling the API endpoint. But this endpoint is the most economical of this hosted model without any compromises in throughput and quality.