OPEN LINK ↗
// 59m agoBENCHMARK RESULT
Claude Fable 5.2 Opens Five-Point Lead
Claude Fable 5.2 now leads Artificial Analysis’s LLM leaderboard with a 65 Intelligence Index score, five points ahead of Gemini 4 Carbon. The result follows Anthropic’s November 9 release and puts Fable’s max-effort configuration at the frontier for coding and difficult reasoning.
// ANALYSIS
Fable 5.2 is the new benchmark king, but the headline margin comes with caveats: it is a weighted index, and reasoning configuration matters as much as the model.
- –Max-effort Fable scores 65, while high effort falls to 62, so production defaults may not reproduce the top-line result.
- –Fable leads on HLE-Diamond, CritPt, and FrontierCode, while Gemini 4 Carbon remains ahead on AutomationBench and Terminal-Bench 4.0.
- –Max-effort tasks can be expensive and slow, reaching roughly $27 per task and 44 minutes in the tracked comparison.
- –Developers should route long, high-value coding and research jobs to Fable, then validate performance against real workloads before changing defaults.
// TAGS
claude-fable-5-2llmbenchmarkevaluationreasoningai-coding
DISCOVERED
59m ago
2026-11-10
PUBLISHED
1h ago
2026-11-10
RELEVANCE
9/ 10
AUTHOR
theo