Add Opus 4.7, GPT 5.5, GPT Image 2, improve model update command

[bubacoder]

Summary by CodeRabbit

• New Features

  • Added support for Claude Opus 4.7, GPT-5.5, and multiple new model variants including DeepSeek v4-pro and Grok 4.20.

• Updates

  • Updated AI model routing and fallback configuration for improved availability.

[coderabbitai]

Walkthrough

This PR updates AI model configurations by revising the LiteLLM update policy documentation to reflect a new add/keep/remove strategy, and modifies the LiteLLM model configuration to add new model mappings and update existing ones across multiple providers (Anthropic, OpenAI, OpenRouter).

Changes

Cohort / File(s)	Summary
Documentation .claude/commands/update-ai-models.md	Updated AI model configuration policy documentation, renaming from "updating LiteLLM config" to broader "AI model configurations." Revised the per-model update rules from Update to Add, introduced explicit "Keep (latest)" and "Keep (previous)" retention rules, and tightened removal criteria to entries two or more versions behind.
LiteLLM Model Configuration docker/ai/litellm/config/config.yaml	Added new model mappings for Anthropic claude-opus-4-7, OpenAI gpt-5.4-mini and gpt-5.5, OpenRouter gpt-5.4, deepseek-v4-pro, grok-4.20, and gpt-5.4-image-2. Updated existing model versions and renamed entries across providers. Adjusted router_settings.fallbacks routing rules to point gpt-5.5 and gpt-5.4-mini to gpt-5.4.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~12 minutes

Possibly related PRs

• Update LiteLLM models #288 — Modifies the same LiteLLM config file and update documentation with overlapping model configuration changes.
• Update LiteLLM model configurations and Ollama pull script to use latest model versions #312 — Updates the same LiteLLM config file with overlapping model entries (gpt-5.4, gpt-5.4-mini) and router fallback mappings.
• Configure OpenAI provider and GPT-5 models #174 — Modifies LiteLLM configuration to add/change OpenAI and GPT model entries with fallback routing updates.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately captures the main changes: adding new AI models (Opus 4.7, GPT 5.5, GPT Image 2) and improving the model update command documentation with revised versioning policies.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feature/update-ai-models

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🧹 Nitpick comments (2)

docker/ai/litellm/config/config.yaml (1)

28-31: Consider adding a claude-opus-4-7 → claude-opus-4-6 fallback for parity.

You added a gpt-5.5 → gpt-5.4 fallback alongside the new gpt-5.5 entry, but the analogous Opus 4.7 → 4.6 fallback isn't defined. If the rationale for the GPT fallback (handle outages / capacity errors on the freshly released model by degrading to the previous version) applies here too, this looks asymmetrical.

♻️ Suggested addition

 router_settings:
   fallbacks:
+    - claude-opus-4-7:
+        - claude-opus-4-6
     - gpt-5.5:
         - gpt-5.4
     - gpt-5.4-mini:
         - gpt-5.4
     - ollama-mac-mistral:
         - ollama-local-phi

Also applies to: 158-165

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@docker/ai/litellm/config/config.yaml` around lines 28 - 31, Add a parallel
fallback entry for Claude Opus by defining a second config block with
model_name: claude-opus-4-6 and litellm_params.model set to
anthropic/claude-opus-4-6 (and the same api_key reference) so the setup mirrors
the gpt-5.5 → gpt-5.4 fallback; update the existing claude-opus-4-7
configuration areas (the model_name and litellm_params blocks) to include this
new claude-opus-4-6 entry wherever the gpt fallback was added (ensure keys match
the existing pattern: model_name and litellm_params.model/api_key).

.claude/commands/update-ai-models.md (1)

88-93: Step 4 verification only covers LiteLLM, not the Ollama script.

Step 3 writes both docker/ai/litellm/config/config.yaml and scripts/get-offline-data-ollama.sh, but Step 4 only re-reads the LiteLLM config. Consider also verifying the Ollama script (or just git diff from the repo root to give the user a single review of all written files), so a botched tee heredoc on either file is caught.

📝 Suggested addition

 ## Step 4: Verify

-Confirm the LiteLLM config was written correctly:
+Confirm both files were written correctly:
 ```bash
 docker exec litellm cat /app/config.yaml
+git -C "$(git rev-parse --show-toplevel)" diff -- \
+  docker/ai/litellm/config/config.yaml \
+  scripts/get-offline-data-ollama.sh

</details>

<details>
<summary>🤖 Prompt for AI Agents</summary>

Verify each finding against the current code and only fix it if needed.

In @.claude/commands/update-ai-models.md around lines 88 - 93, Step 4 only
re-reads the LiteLLM config; also add verification for the Ollama script
(scripts/get-offline-data-ollama.sh) or a single repo-root git diff that shows
both files were written; update the verification step to run the existing docker
exec litellm cat /app/config.yaml check and additionally run a git diff (against
docker/ai/litellm/config/config.yaml and scripts/get-offline-data-ollama.sh) so
a failed heredoc/tee on either file is detected.

</details>

</blockquote></details>

</blockquote></details>

<details>
<summary>🤖 Prompt for all review comments with AI agents</summary>

Verify each finding against the current code and only fix it if needed.

Inline comments:
In @docker/ai/litellm/config/config.yaml:

• Around line 86-90: The duplicate model_name "gpt-5.4" in the OpenRouter entry
  causes LiteLLM to treat both providers as one load-balanced deployment; to fix,
  change the OpenRouter entry's model_name (e.g., to "gpt-5.4-or") and keep its
  litellm_params.model set to "openrouter/openai/gpt-5.4", then add a fallback
  rule in router_settings.fallbacks that maps requests for "gpt-5.4" to use
  "gpt-5.4-or" only as a fallback (instead of relying on model_list load
  balancing) so billing/latency/rate-limit boundaries remain separate.

---

Nitpick comments:
In @.claude/commands/update-ai-models.md:

• Around line 88-93: Step 4 only re-reads the LiteLLM config; also add
  verification for the Ollama script (scripts/get-offline-data-ollama.sh) or a
  single repo-root git diff that shows both files were written; update the
  verification step to run the existing docker exec litellm cat /app/config.yaml
  check and additionally run a git diff (against
  docker/ai/litellm/config/config.yaml and scripts/get-offline-data-ollama.sh) so
  a failed heredoc/tee on either file is detected.

In @docker/ai/litellm/config/config.yaml:

• Around line 28-31: Add a parallel fallback entry for Claude Opus by defining a
  second config block with model_name: claude-opus-4-6 and litellm_params.model
  set to anthropic/claude-opus-4-6 (and the same api_key reference) so the setup
  mirrors the gpt-5.5 → gpt-5.4 fallback; update the existing claude-opus-4-7
  configuration areas (the model_name and litellm_params blocks) to include this
  new claude-opus-4-6 entry wherever the gpt fallback was added (ensure keys match
  the existing pattern: model_name and litellm_params.model/api_key).

</details>

<details>
<summary>🪄 Autofix (Beta)</summary>

Fix all unresolved CodeRabbit comments on this PR:

- [ ] <!-- {"checkboxId": "4b0d0e0a-96d7-4f10-b296-3a18ea78f0b9"} --> Push a commit to this branch (recommended)
- [ ] <!-- {"checkboxId": "ff5b1114-7d8c-49e6-8ac1-43f82af23a33"} --> Create a new PR with the fixes

</details>

---

<details>
<summary>ℹ️ Review info</summary>

<details>
<summary>⚙️ Run configuration</summary>

**Configuration used**: Path: .coderabbit.yaml

**Review profile**: CHILL

**Plan**: Pro

**Run ID**: `72beacbe-2922-491d-add6-0ff47b083619`

</details>

<details>
<summary>📥 Commits</summary>

Reviewing files that changed from the base of the PR and between b8d3a29af0b7fa25fe77f8b33e18f997130d4c45 and 7c13693eebeeef7a763ef280562a9302a95a85ba.

</details>

<details>
<summary>📒 Files selected for processing (2)</summary>

* `.claude/commands/update-ai-models.md`
* `docker/ai/litellm/config/config.yaml`

</details>

</details>

<!-- This is an auto-generated comment by CodeRabbit for review status -->