Claude Onigiri EAP surfaces in code

// 1h agoNEWS

Claude Onigiri EAP surfaces in code

An unreleased Claude checkpoint, claude-onigiri-eap-2, appeared in Artificial Analysis frontend code with low, medium, high, and xhigh reasoning variants. Its Opus 4.8 fallback suggests a possible Fable 5.5 connection, but Anthropic has not confirmed the model.

// ANALYSIS

This looks like credible pre-release plumbing, not a launch—useful evidence that Anthropic may be testing another frontier model, but far too little to identify it confidently.

–Five Onigiri identifiers shared the same Opus 4.8 fallback configuration
–The reasoning-level variants point toward adaptive effort controls for demanding coding and knowledge-work tasks
–The code entries contain no release date, suggesting an internal or early-access build rather than a public API release
–Fable 5.5 remains speculation; the fallback pattern is suggestive but not unique
–Developers should wait for official model IDs, pricing, documentation, and benchmark results before planning integrations

// TAGS

claude-onigiri-eapllmreasoningevaluationinference

DISCOVERED

1h ago

2026-10-16

PUBLISHED

1h ago

2026-10-16

RELEVANCE

9/ 10

AUTHOR

WorldofAI

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

NEWS1h ago

Claude Gabagool EAP Fuels Opus 6 Speculation

WorldofAI reports that an unreleased Claude checkpoint named Gabagool EAP appeared alongside Onigiri in Artificial Analysis code, fueling speculation about a future Opus 6 or upgraded Opus model. Anthropic has not publicly confirmed the checkpoint; its Opus page documents currently released models and availability.[Claude Opus](https://www.anthropic.com/claude/opus)

VIDEO1h ago

FingerprintJS fingerprints browsers across incognito sessions

Better Stack’s video demonstrates how FingerprintJS combines browser attributes into a hashed visitor ID that can remain stable across private sessions and cleared site data. The open-source client-side library is useful for fraud and abuse detection, but its accuracy is lower and fingerprints can be spoofed or disrupted by privacy-focused browsers.

RESEARCH2h ago

Disagreement-Guided Routing Makes Weak Models Defer

University of Pennsylvania researchers introduce a ground-truth-free routing method that uses disagreement between weak and strong language models to decide when escalation is worthwhile. Tests across agentic environments and reasoning benchmarks show improved cost–accuracy trade-offs over uncertainty and consistency baselines.