fix(ai-gateway): update litellm version to 1.100.0

[Gary Henderson (Gary-H9)]

Summary

• Updates the active LiteLLM image in contrib/docker-compose-ai-gateway.yml from v1.98.0 to v1.100.0.
• v1.100.0 is the latest published non-draft, non-prerelease LiteLLM release as at 2026-09-09; no newer stable release exists as of this run's cutoff.
• Catches the local Compose companion up to the already-merged Modernisation Platform first-stage rollout in modernisation-platform-environments#19030, which moved development/test to 1.100.0 without a paired Compose update at the time.
• Paired with the pre-production/production promotion in modernisation-platform-environments#19047, opened in this same rollout run.
• Validation completed: docker compose -f contrib/docker-compose-ai-gateway.yml config -q, git diff --check, ruff format --check, and ruff check . passed; exactly one active LiteLLM image reference uses v1.100.0. All 26 focused AI Gateway client tests (tests/ai_gateway/test_client.py) passed.

Data Platform client compatibility

No breaking changes affecting the Data Platform AI Gateway client were identified between 1.98.0 and 1.100.0. The authoritative latest-main AIGatewayClient was checked against the complete v1.98.0...v1.100.0 upstream range:

• Access groups and organisations: access_group_endpoints.py and organization_endpoints.py are byte-for-byte unchanged since v1.98.0 (source: BerriAI/litellm compare above shows no diff for either file), so GET /v1/access_group and GET /organization/list?org_alias=... retain their paths, methods and access_group_name, access_group_id, access_model_names, organization_alias, organization_id fields.
• Teams: team_endpoints.py is likewise unchanged over the full range. POST /team/new, POST /team/delete, GET /team/info?team_id=..., and POST /team/update retain the client's payloads and the team_id, team_info, and access_group_ids response fields. The client's budget_duration: "monthly" value is unaffected.
• Virtual keys: key_management_endpoints.py is unchanged over the full range. POST /key/generate, POST /key/{key}/regenerate, POST /key/delete, GET /key/info?key=..., GET /key/list (with team_id, return_full_object, size, page), and POST /key/update retain the bodies and key, keys, models, total_pages fields consumed by the client.
• Models: proxy_server.py is unchanged over the full range, so GET /v1/model/info retains the data response field consumed by list_models_v1_info.
• Authentication and errors: the client continues to send the master key as a bearer token and raises on non-success responses; no auth or error-shape change was introduced in this range. The focused client contract suite passed all 26 tests, including authentication, error handling and /key/list pagination.

This review builds on the previously merged 1.97.0-to-1.98.0 assessment and the 1.98.0-to-1.99.1 assessment (reused; that PR was closed unmerged but its source-level review remains valid evidence), extended here with a fresh 1.99.1-to-1.100.0 check. The application uses HTTP through pinned httpx==0.28.1, not the LiteLLM Python package.

Other breaking changes

• No breaking-change entry applicable to this deployment was identified in the v1.99.0, v1.99.1, or v1.100.0 release notes.
• The tagged schema.prisma adds nullable/defaulted columns and new tables only (for example LiteLLM_ModelAccessGroupBudgetTable, LiteLLM_BudgetWindowSpend, LiteLLM_ProxyWorkerHeartbeat, LiteLLM_DailyGuardrailUsageUnits, and additive gateway_injected_caching_savings_spend/shadow-eval columns); the supplied SQL migrations remain additive across the full 1.98.0...1.100.0 range. Evidence: BerriAI/litellm schema diff.
• v1.99.1 is a Docker-only release (no PyPI package); this Compose file already deploys ghcr.io/berriai/litellm images, so it is unaffected.
• The Data Platform App does not import the LiteLLM Python package, so no Python client-library coupling is introduced by this range.

Release notes

• LiteLLM 1.100.0 release notes
• Full upstream comparison v1.98.0...v1.100.0

[github-actions]

Dependency Review

✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.

Scanned Files

None

[Copilot]

🟢 Approval recommended

The change is a straightforward, verified image-tag bump in a single Compose file with no additional code-path impact.

Pull request overview

Updates the local AI Gateway Docker Compose setup to run the current stable LiteLLM container image, keeping developer/local-compose parity with the environment rollout.

Changes:

• Bumps the ghcr.io/berriai/litellm image tag in the AI Gateway Compose file from v1.98.0 to v1.100.0.

File summaries

File	Description
contrib/docker-compose-ai-gateway.yml	Updates the LiteLLM container image tag used by the local AI Gateway Compose stack.

Review details

• Files reviewed: 1/1 changed files
• Comments generated: 0
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Gary Henderson (Gary-H9)]

Closing this automation-created PR because the current run is a pre-production/production promotion. The rollout state machine requires promotion to take precedence and explicitly defers a new Data Platform App image update during the promotion run.