Hermes Agent Adds Claude Fable 5.2

// 38m agoPRODUCT UPDATE

Hermes Agent Adds Claude Fable 5.2

Nous Research has added Anthropic’s Claude Fable 5.2 to Hermes Agent through Nous Portal. The model offers a 1M-token context window, 128K-token outputs, and adaptive thinking for long-running agent workflows.

// ANALYSIS

This makes Hermes more compelling as an open, model-agnostic agent harness for testing frontier models against real tools and workflows.

–Hermes’ model catalog lists Fable 5.2 alongside models from OpenAI, Google, DeepSeek, Qwen, and others, enabling practical model comparisons. [Model catalog](https://github.com/NousResearch/hermes-agent/blob/main/website/static/api/model-catalog.json)
–The model’s long context and high output ceiling fit multistep research, analysis, and document-generation tasks.
–At $8 input and $40 output per million tokens, Fable 5.2 is premium-priced; developers should reserve it for high-value jobs.
–Hermes’ open-source, self-installable runtime gives developers a useful environment for evaluating agent behavior beyond chat interfaces. [Hermes Agent docs](https://hermes-agent.nousresearch.com/docs)

// TAGS

hermes-agentclaude-fable-5-2agenttool-uselong-contextreasoningopen-source

DISCOVERED

38m ago

2026-11-09

PUBLISHED

1h ago

2026-11-09

RELEVANCE

9/ 10

AUTHOR

NousResearch

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

RESEARCH22m ago

Anthropic Paper Finds Weak-Coupling Transition First Order

An Anthropic-contributed Hexagon preprint reports that the Hawking–Page-like transition in weakly coupled N=4 super Yang–Mills remains first order, based on a negative quartic Polyakov-loop coefficient. The AI-assisted calculation was checked against known cases, but remains an unreviewed preprint.

RESEARCH22m ago

Claude Derives Sharp Large-c Bootstrap Bound

Anthropic’s primarily AI-generated Hexagon preprint reports an exact asymptotic slope, ΛV ≈ 9.1266, for spinless modular-bootstrap bounds on 2D CFT spectral gaps. The result narrows a previously broad range while remaining limited to a restricted, computer-assisted method. [Hexagon preprint](https://hexagonmath.org/2611.00174v1)

MODEL22m ago

Claude Fable 5.2 Formalizes Fermat With 1,000 Agents

Anthropic says 1,000 Claude Fable 5.2 agents formalized Fermat’s Last Theorem in Lean in about 30 hours, using shared SQL state, Git branches, automated grading, and layered safety controls. The run used one-third as many output tokens and produced a proof one-seventh the length of an earlier 11-day effort.