docs: refresh retired model ids across the integration guides

[MVS-source]

What

Ten model ids cited across v3/integrations/*.mdx were no longer served. This replaces all of them. 13 files, 76 insertions and 76 deletions, so it is a pure substitution with no structural change.

Dead id	Occurrences	What was wrong	Replacement
anthropic/claude-sonnet-4-5	48, in 13 of 24 pages	model retired	anthropic/claude-sonnet-latest
anthropic/claude-opus-4.7	3, claude-code	typo, dot instead of dash	anthropic/claude-opus-4-7
cohere/command-r-plus	5	date suffix missing	cohere/command-a-03-2025
cohere/command-r	1	date suffix missing	cohere/command-r-08-2024
meta/llama-3-70b	2	there is no meta/ vendor	scaleway/llama-3.3-70b-instruct
meta/llama-3-8b	1	there is no meta/ vendor	ionos/meta-llama/Meta-Llama-3.1-8B-Instruct
openai/dall-e-3	2	absent from /v3/images/models too	openai/gpt-image-2
openai/gpt-5.1-codex, -mini, -max	5, codex-cli	only gpt-5.3-codex exists under openai/	openai/gpt-5.3-codex plus the Azure-hosted variants

anthropic/claude-sonnet-4-5 was our house example. Where an alias exists it is now used instead of a pinned version, so the same ids will not rot again: anthropic/claude-sonnet-latest resolves to claude-sonnet-5 today and will follow.

For Llama, the ids now name their host because that is how the catalogue exposes them. The headings were adjusted so a reader is not surprised by a scaleway/ or ionos/ prefix under a Meta heading.

Verification

Every replacement was checked with a real API call, not just for presence in the catalogue, because the catalogue can list a model it will not serve. All seven chat replacements returned 200, openai/gpt-image-2 returned 200 on /v3/images/generations, and the codex ids returned 200 on /v3/responses.

After the sweep, a pass over all 24 pages against the live chat, image and embeddings catalogues finds zero dead ids. The only remaining matches are three false positives that are not model ids at all: openai/resources/chat/completions (a TypeScript import path), oauth/google/callback (a URL) and github.com/openai/codex (a repo URL).

Human-readable labels sitting next to the ids were realigned too, otherwise the page would say "Command R+" next to a Command A id: three of those in continue-dev and openclaw.

One real bug this turned up

The auth-check curl in the Codex CLI guide sent a Codex model to /chat/completions. Codex models are served on /responses only, so that check returned:

Model(s) do not support /chat/completions: 'openai/gpt-5.3-codex' (supports: /responses)

A 400, on the page whose whole purpose at that point is "check whether your key works". The check now uses a plain chat model, with a sentence saying why. The rest of the guide was already correct, it sets wire_api = "responses".

Note on CI

Execution Tests is red here as it is on every PR in this repo since around 11 August. It dies in pytest_sessionstart on tests/helpers/api.py getting 401 Unauthorized from https://staging-api.edenai.run/v2/user/custom_token/. Not related to this change. Same on #77, #90 through #99.

[mintlify]

Preview deployment for your docs. Learn more about Mintlify Previews.

Project	Status	Preview	Updated (UTC)
edenai	🟢 Ready	View Preview	Aug 28, 2026, 9:56 AM

💡 Tip: Enable Workflows to automatically generate PRs for you.

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: cc60a4a4-abc1-4083-84dc-ca1e5cd58780

---

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