Netlify Adds Fable 5.1 With 75% Cheaper Cache Reads.

Netlify has added Fable 5.1 to its AI Gateway and Agent Runners, giving developers access to Anthropic’s latest model without separately configuring provider credentials. The headline improvement is economic: Netlify says cached input reads cost 75% less than they did with Fable 5.

That reduction is particularly relevant for coding agents, which may repeatedly inspect the same repository files while planning changes, answering questions, or carrying out a longer task. Each cached read is only one part of an agent’s total model bill, but the savings can compound across repositories and repeated runs.

Netlify announced the integration on LinkedIn on September 1. Its official changelog says the model is available through both AI Gateway and Agent Runners with no configuration required.

Developers can call Anthropic’s Claude Fable 5.1 model from Netlify Functions through AI Gateway. Netlify says teams can use the Anthropic SDK directly, while its gateway manages authentication behind the scenes. The model identifier in the company’s example is claude-fable-5-1.

AI Gateway also provides Netlify’s surrounding infrastructure for model calls, including caching and rate limiting. That allows a team to keep AI-backed application code inside its existing Netlify deployment rather than wiring up provider authentication separately for every function or environment. The company’s AI Gateway documentation provides the implementation details.

The second route is Agent Runners, Netlify’s system for assigning development work to agents within a project workflow. Users can select Fable 5.1 when starting an agent run. This follows Netlify’s recent work on clarifying questions for Agent Runners, which lets an agent resolve unclear instructions before it begins changing code.

Netlify has also introduced an Ask mode that lets agents inspect an application without immediately modifying it. Together, those controls and a model with cheaper cache reads point toward more iterative agent workflows, where an agent can examine a repository, ask for missing information, and revisit relevant context throughout a task.

Prompt caching avoids processing unchanged context from scratch on every request. In software work, that context can include source files, configuration, documentation, and earlier instructions. An agent operating across several steps may need to reference those materials repeatedly.

Netlify’s claim is specifically that cache reads are 75% cheaper compared with Fable 5. It does not mean every Fable 5.1 request or complete agent run will cost 75% less. Final spending will still depend on factors such as how much context is cached, how often the agent reuses it, and the volume of uncached input and generated output.

For small development teams, the practical impact should be clearest on repository-heavy tasks. A short, isolated request may see little benefit. Longer jobs that repeatedly inspect the same codebase have more opportunities to reuse cached context and therefore capture the lower read price.

The addition also broadens model choice inside Netlify. Fable 5.1 arrived one day before the platform added Gemini 3.8 Flash across Functions and Agent Runners. Teams can now choose models according to the task, latency requirements, and cost profile without rebuilding the surrounding deployment workflow each time.

What is Netlify Fable 5.1?

Fable 5.1 is an Anthropic Claude model now supported by Netlify’s AI Gateway and Agent Runners. Developers can call it from Netlify Functions or select it for an agent run, while Netlify handles gateway authentication and related infrastructure.

How much cheaper are Fable 5.1 cache reads?

Netlify says Fable 5.1 cache reads are 75% cheaper than cache reads with Fable 5. The reduction applies to cached reads, not necessarily the entire request or total cost of an Agent Runner task.

When and where is Fable 5.1 available on Netlify?

Netlify announced availability on September 1, 2026. The model is available through AI Gateway and Agent Runners, with Netlify describing the integration as requiring zero configuration.