Thursday, November 19, 2026

NEAR AI Adds Claude Fable 5.2 to Cloud

Photorealistic cloud hub with Claude Fable 5.2 UI, NEAR Protocol emblem, and soft editorial lighting in neutral tones.

NEAR AI Adds Claude Fable 5.2 to Cloud

NEAR AI has added Anthropic’s Claude Fable 5.2 to NEAR AI Cloud, giving developers access to the frontier model through the platform’s API and browser-based playground. Anthropic released Fable 5.2 on November 9 for demanding reasoning, coding and long-horizon agentic workflows. The integration expands NEAR AI’s model catalog without moving Fable’s underlying inference onto NEAR-operated decentralized compute.

According to the official NEAR AI model page, Fable 5.2 supports a 1-million-token context window, image input and a maximum output of 128,000 tokens. NEAR AI lists pricing at $8 per million input tokens, $40 per million output tokens and $0.20 per million cached input tokens. Developers can call the model through NEAR AI’s Anthropic-compatible Messages API using the anthropic/claude-fable-5-2 model identifier.

NEAR Routes Requests While Anthropic Runs Inference

NEAR AI classifies Fable 5.2 as an Incognito model. Under that configuration, NEAR’s TEE-based router uses shared provider credentials so the upstream provider does not receive the user’s NEAR AI account credentials, but the provider still processes the prompts on its own infrastructure. Fable 5.2 does not receive the hardware-attested confidential-inference guarantees available to NEAR AI’s Private models.

That distinction is particularly relevant as NEAR expands access to different privacy tiers through its Gateway for Claude Code, Codex and OpenCode. Private models can execute inside secure hardware enclaves, whereas Incognito models such as Fable route requests to external providers. The broader architecture also includes verifiable confidential inference for supported AI agents and models. Model availability through the same cloud interface does not mean every model receives the same privacy or execution guarantees.

Anthropic designed Fable 5.2 around adaptive thinking, which is always enabled. Developers can use the effort parameter to control reasoning depth, latency and token consumption, with high as the default setting. Adaptive thinking changes how computational effort is allocated to a task rather than expanding the model’s underlying context window.

Anthropic also says Fable 5.2 can generate output up to 30% faster than Fable 5.1 while reducing standard API prices from $10/$50 to $8/$40 per million input and output tokens. That percentage is an Anthropic model-level performance claim rather than a benchmark specifically measured on NEAR AI Cloud. The previously cited increase from roughly 83 to 98 tokens per second would equal about 18%, not 30%, and is not sufficiently supported by the available primary documentation.

Integration Extends NEAR’s Agent Infrastructure

The large context window is designed for workloads that require substantial persistent context, including long-running coding sessions, multistep research and agents working through extended tasks. Anthropic describes Fable 5.2 as intended for agent sessions lasting hours and complex work carried through to finished outputs. A 1-million-token window increases the amount of context available to an agent, but it does not by itself demonstrate autonomous execution or successful long-running production use.

NEAR is simultaneously building infrastructure around other parts of agent execution. Its ecosystem now includes private USDC payments designed for agentic commerce, creating a separate settlement layer for software capable of purchasing services or interacting with financial infrastructure. Inference, agent authorization and blockchain settlement remain distinct technical layers even when they are combined within a broader agent stack.

The approach also differs from decentralized GPU networks that provide the underlying hardware for AI workloads. Projects such as io.net are pursuing distributed GPU infrastructure optimized for AI training and inference, whereas Fable 5.2 requests made through NEAR AI ultimately execute on Anthropic’s provider infrastructure. Access to a model through a blockchain-associated AI platform should therefore not be confused with decentralized execution of the model itself.

Reuters reported that Anthropic released Fable 5.2 alongside Claude Mythos 5.2 on November 9 as competition among frontier-model providers intensified. Fable is also available through Anthropic’s own API and major cloud platforms including AWS, Google Cloud and Microsoft Foundry. NEAR AI is therefore providing an additional access and routing layer for an externally hosted model rather than exclusive infrastructure for Fable 5.2.

The measurable development is narrower than a shift toward decentralized frontier-model execution: developers can now access Claude Fable 5.2 through NEAR AI Cloud with its full 1-million-token context window and standard Anthropic pricing. Whether the integration develops into meaningful adoption will depend on observable API usage and recurring agent workloads rather than model availability alone.

Satoshipick
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.