Claude Opus 5 tops key AI benchmarks while undercutting Fable 5 on cost

Anthropic’s Claude Opus 5 scores 61 on the Artificial Analysis Intelligence Index, putting it just ahead of Claude Fable 5 at 60 and GPT-5.6 Sol at 59. The index combines nine tests covering knowledge work, coding, scientific reasoning, and factual accuracy. The takeaway is clear: Opus 5 is at the front of the pack, but the margin is slim enough that no single frontier model has a decisive lead.

In coding, Opus 5 at “xhigh” paired with Claude Code shares first place on the Artificial Analysis Coding Index. On Terminal-Bench v2.1, Opus 5 scored 89 percent at “max,” matching the previous leader GPT-5.6 Sol. Vals.ai testing found that the “high” tier delivered the best coding score among listed tiers at 89.8 percent, while higher tiers produced more complex solutions that more often contained errors.

The average Intelligence Index task costs $2.03 with Opus 5, below Claude Fable 5 with fallback at $2.75, though above Opus 4.8 at $1.80 and Sonnet 5 at $1.53. At “high” and “xhigh,” Opus 5 beats both Opus 4.8 and Sonnet 5 while maintaining lower costs. On AA-Briefcase, Opus 5 at “high” costs $10.41 per task, less than half of Fable 5’s $22.30, while still beating Fable 5 in the Elo ranking.

Factual accuracy remains a weak spot for Opus 5. On AA-Omniscience, it improved by 7 points over Opus 4.8 but still trails Fable 5. Because Opus 5 answers more often when uncertain, its hallucination rate rises 14 points to 50 percent, making verification important for high-stakes work.

Opus 5 performs especially well on AA-Briefcase, which evaluates office-style tasks such as research reports, presentations, and spreadsheet analysis. At max reasoning, it reaches an Elo of 1720, 146 points ahead of Claude Fable 5 at 1574, and its three highest tiers take the top three spots. The model’s strongest gains are in analytical quality, though presentation quality trails GPT-5.6 Sol at “max.”