Fable 5.2 Retakes #1 LLM Spot Amid Anthropic's IPO Roadshow
Story 20, from Latent Space
Anthropic's Fable 5.2 has reclaimed the top LLM position, outperforming Gemini 4 Carbon and Opus 5.5 in benchmarks. It comes with a 20% price cut per token but higher computational demands.
Key points
- Fable 5.2 is now the top LLM, surpassing Gemini 4 Carbon and Opus 5.5 in most benchmarks.
- Pricing is $8 per million input tokens and $40 per million output tokens, a 20% reduction from Fable 5.1.
- Fable 5.2 generates more test cases and uses subagents for verification, contributing to its performance edge.
- Effort levels like High and XHigh show better efficiency than Max, which underperforms in some benchmarks.
- Subscription access is limited, with Ultra and Team Premium users getting more resets.
- Fable 5.2 is available on multiple platforms including Cursor, GitHub Copilot, and Perplexity.
Why it matters
As a software engineer, you may want to evaluate Fable 5.2 for its performance in complex tasks and cost efficiency, especially if you're working on projects that require high reasoning and code generation capabilities.
Counterpoint
Fable 5.2's Max effort level underperforms in some benchmarks, and its higher computational demands may offset the cost savings, making it less suitable for certain workloads.
Sources
This summary was written by a language model and may contain errors. Read the original for the full story.