AI Product Updates, July 20: GitHub Code Quality Goes Paid, GrowthBook Ships 5.0 (2026)

The short version

GitHub Code Quality left public preview and became a billed product on Enterprise Cloud and Team. GrowthBook shipped version 5.0 with agent skills, an in-app assistant, and a CLI that covers its API. Anthropic changed how paid users consume Claude Fable 5 after its introductory promotion ended. SAP used a July 20 post to document a broader Q2 bundle, including general-availability agents for manufacturing, code migration, and mobile work.

Product	Releasing entity	Release date	What changed	Availability
GitHub Code Quality	GitHub	July 20, 2026	Public preview became GA, adding organization dashboards, test-coverage metrics, ruleset quality gates, and APIs for enablement and findings. 1	Enterprise Cloud and Team. $10 per active committer per month, plus metered AI work and CodeQL compute. Not on Enterprise Server at launch.
GrowthBook 5.0	GrowthBook	July 20, 2026	The release adds an agent experience with official skills, an in-app assistant, and a full-API CLI. Its launch week also stages an AI Visual Editor, Product Analytics GA, governance, and faster experiments. 2	Live on GrowthBook Cloud and available for self-hosted deployments. Pricing changes are not stated in the announcement.
Claude Fable 5 plan access	Anthropic	July 20, 2026	A promotion that included up to 50% of weekly usage at no extra cost ended July 19. Fable 5 remains available, but plan entitlements now differ. 3	All paid plans, not Free. Max and premium Team or Enterprise seats include up to 50% of weekly limits; Pro and standard seats use usage credits.
SAP Business AI Q2 release bundle	SAP	April-June 2026; recap published July 20	A retrospective covering Joule Work mobile, manufacturing and code-migration agents, SAP AI Core controls, and other enterprise AI features. 4	Mixed: general availability, beta, and early-adopter programs. It is a Q2 release recap, not a same-day launch list.

GitHub turns Code Quality into a billed product

GitHub Code Quality is now generally available on GitHub Enterprise Cloud and GitHub Team. The service combines deterministic CodeQL analysis with AI-assisted detection for maintainability and reliability issues in pull requests, while Copilot Autofix suggests fixes for review before merge. GitHub says more than 10,000 enterprises used it during public preview and that its own engineering organization resolves 67.3% of findings before merge. The percentage is a GitHub-reported internal result, not an independent benchmark. 1

The product added organization-wide enablement, repository dashboards, Cobertura XML coverage metrics on pull requests, quality gates through GitHub rulesets, and APIs for repository enablement and findings. Those additions make it a governance surface for teams using AI coding tools, rather than only another scan attached to an individual repository.

The commercial change is the part to check first. Public preview has ended; billing starts automatically at GA. The new price is $10 per active committer per month, where an active committer pushed to an enabled repository within the last 90 days. GitHub also charges for AI-powered work and for the compute used by deterministic CodeQL analysis. The announcement does not give a prior public-preview price. Enterprise Server is excluded at launch.

GrowthBook 5.0 gives agents a direct product surface

GrowthBook 5.0 is live on Cloud and available for self-hosted deployments. The Monday launch material names three agent-facing pieces: official skills, a new in-app assistant, and a CLI covering the platform's API. Agents can operate GrowthBook to ship flags, launch experiments, analyze results, and clean up after them. 2

The release is organized as a week-long rollout rather than one feature dump. GrowthBook lists an AI Visual Editor for Tuesday, Product Analytics moving to GA on Wednesday, feature-flag governance on Thursday, and faster experiment setup and queries on Friday. Those items are part of the 5.0 launch plan, but the page does not say that every one was live on July 20. The practical availability is clearer: the platform is already running on GrowthBook Cloud, and the release is available to self-hosted users.

The announcement does not give a new price or a previous-versus-current comparison. Teams evaluating it should therefore separate the available deployment choices from the still-unpriced packaging details.

Anthropic changes how paid users consume Fable 5

Anthropic's previous Fable 5 promotion ended at 11:59:59 PM PT on July 19. Starting July 20, Fable 5 is still available on every paid Claude plan, but it is no longer included in the same way for every seat. It is not available on the Free plan. 3

Max plans, premium Team seats, and premium seats on legacy seat-based Enterprise plans include Fable 5 within regular weekly limits, up to 50% of the weekly limit at no extra cost. Further use can draw on usage credits.
Pro plans, standard Team seats, and standard seats on legacy seat-based Enterprise plans use usage credits from the start. Eligible Pro and standard Team seats receive a one-time credit under the transition promotion; standard legacy Enterprise seats do not qualify for that credit.
Usage-based Enterprise and Claude API access are billed at standard API rates. The page directs customers to Anthropic's current pricing page rather than publishing a new Fable 5 rate in this update.

Fable 5 is selectable across Claude web, mobile, and desktop, plus Cowork, Claude Code, Claude Design, Microsoft 365, and Claude Tag. Claude Code needs version 2.1.170 or later, and Cowork needs the latest Claude Desktop. For procurement and engineering teams, the important change is the seat-to-credits boundary, not a model-version change.

SAP's July 20 post is a Q2 enterprise AI catch-up

SAP's release-highlights post is dated July 20, but it explicitly covers offerings released from April 1 through June 30. It should be read as a status recap, not evidence that all of these features launched yesterday. Availability is mixed across the bundle. 4

The general-availability items include the Joule Work mobile app, which lets employees query SAP systems and complete work from iPhone, iPad, and other mobile devices; AI-assisted production engineering in SAP Digital Manufacturing, which analyzes error logs and can generate resolution instructions and script tasks; and the SAP S/4HANA custom code migration agent, which applies high-confidence fixes automatically while leaving lower-confidence proposals for developer review. SAP reports the operational improvements; the post does not provide independent validation.

The same recap places Joule Studio and Voice for Joule in SAP's Early Adopter Care program, with registrations closed for those entries. It also documents SAP AI Core capabilities including a batch API for asynchronous high-volume model workloads and an inference observability service that records prompts, responses, feedback, labels, and filters. For teams buying enterprise AI, these controls matter as much as the agent demos, but the access status varies by product.

What to check next

GitHub administrators should inventory repositories with Code Quality enabled before the new per-committer and usage charges accumulate. Enterprise Server is still outside the launch scope.
Growth teams can test GrowthBook 5.0 in Cloud or self-host it now, but should confirm which launch-week features have reached their tenant before planning around them.
Claude admins should map Fable 5 access to seat type and decide whether usage credits are acceptable for Pro or standard Team users. Free users cannot use the model.
SAP customers should treat the Q2 list as a mixed-availability catalog and verify the individual Discovery Center or help page before assuming GA access.

AI Product Updates, July 20: GitHub Code Quality Goes Paid, GrowthBook Ships 5.0