8 changes verified July 26, 2026

Know what will break before it ships.

A source-backed compatibility database for AI APIs. Find model shutdowns, exact error symptoms, affected identifiers, and the migration fix before production finds them for you.

Official sources monitored by cron
Exact model and parameter IDs
Visible verification dates

Stack compatibility planLocal scan

Paste model IDs, API parameters, endpoints, or code

Your input stays in this browser. Nothing is uploaded or stored.

8Verified changes

12Platforms monitored

5Upcoming cutoffs

100%Official-source records

Deadline watch

Cutoffs that need action

View deadline calendar →

8 days leftAugust 5, 2026

Claude Opus 4.1 retires on August 5, 2026

Replace the retired model ID with claude-opus-4-8 and regression-test model behavior before the cutoff.

Anthropic · Claude API 13 days leftAugust 10, 2026

Gemini embedding-2-preview shuts down on August 10, 2026

Move embedding requests to the generally available gemini-embedding-2 model.

Google · Gemini API 20 days leftAugust 17, 2026

Imagen 4 API models shut down on August 17, 2026

Migrate image-generation workflows to gemini-3.1-flash-image using Google's current image-generation guidance.

Google · Gemini API 29 days leftAugust 26, 2026

OpenAI Assistants API shuts down on August 26, 2026

Use the official Assistants-to-Conversations migration guide to move state and execution to Responses and Conversations.

OpenAI · OpenAI API

Change ledger

Recently announced or effective

Browse all changes →

GoogleGemini APIhigh

Gemini deprecates temperature, top_p, and top_k

Gemini 3.6 Flash and 3.5 Flash-Lite ignore the sampling parameters; Google says future model generations will return HTTP 400 when they are supplied.

parameter deprecationStrip temperature, top_p, and top_k from Gemini requests and encode response constraints in a clear system instruction.

AnthropicClaude APIhigh

Claude Sonnet 5 changes thinking defaults and token counts

Claude Sonnet 5 enables adaptive thinking by default, removes manual thinking budgets, rejects non-default sampling parameters, and tokenizes the same text into roughly 30% more tokens.

breaking behaviorRemove manual thinking budgets and non-default sampling parameters, rely on adaptive thinking, and use Anthropic's token-counting API to recalibrate limits and cost controls.

GoogleGemini APIcritical

Imagen 4 API models shut down on August 17, 2026

Google is shutting down all three Imagen 4 production model IDs and recommends gemini-3.1-flash-image.

model shutdownMove to gemini-3.1-flash-image and test request shape, output handling, latency, image quality, and safety behavior before the deadline.

AnthropicClaude APIcritical

Claude Opus 4.1 retires on August 5, 2026

Anthropic is retiring Claude Opus 4.1 on the Claude API and recommends Claude Opus 4.8 as its replacement.

model retirementUpdate the model ID to claude-opus-4-8, then rerun prompt, tool-use, latency, and output-quality tests before August 5.

AnthropicClaude APIhigh

Claude Opus 4.7 and later reject non-default sampling parameters

Anthropic has deprecated temperature, top_p, and top_k on Claude Opus 4.7 and later; non-default values return an HTTP 400 error.

breaking behaviorStop sending temperature, top_p, and top_k for affected models. Express tone, variability, and formatting requirements in the system prompt instead.

GoogleGemini APIcritical

Gemini embedding-2-preview shuts down on August 10, 2026

Google is shutting down the preview embedding model and lists gemini-embedding-2 as the replacement.

model shutdownChange the model ID to gemini-embedding-2 and validate vector dimensions, storage, and retrieval quality in a staging index before cutover.

From symptom to fix

A changelog that answers the next question.

Provider announcements tell you what changed. Each record here also maps the affected surface, the observable failure, and the practical migration step.

1Paste a model ID, parameter, code fragment, or error.
2Match it against source-backed provider announcements.
3Open the source receipt and apply the migration fix.

anthropic / request.pyHTTP 400 → valid request

response = client.messages.create(
    model="claude-opus-4-8",
-   temperature=0.2,
-   top_p=0.9,
+   system="Be concise and factual.",
    messages=messages,
)

Coverage