## Claude Opus 5: The New Frontier in AI Performance and Accessibility
Claude Opus 5, released on July 24, marks a significant milestone in AI model development. Priced at $5 per million input tokens—identical to its predecessor Opus 4.8 and exactly half the price of Fable 5—Opus 5 not only matches its predecessor’s cost efficiency but also surpasses Fable 5 on most major benchmarks. This new model is set to become the default choice on Claude Max and the strongest option on Claude Pro, effectively replacing Fable 5 as the go-to model for most subscribers.
### Benchmark Performance
In terms of raw performance, Claude Opus 5 shines brightly:
– **Frontier-Bench v0.1**: Scored 43.3% on this agentic coding evaluation, outperforming Fable 5 (33.7%) and OpenAI’s GPT-5.6 Sol (34.4%).
– **ARC-AGI-3**: Achieved a score of 30.2%, significantly outperforming GPT-5.6 Sol’s 7.8%. This benchmark tests genuine problem-solving abilities through novel puzzles.
– **GDPval-AA v2**: Reached an Elo rating of 1,861, surpassing both Fable 5 (1,747) and GPT-5.6 Sol (1,736) on this knowledge work benchmark.
– **Zapier’s AutomationBench**: Earned a perfect 100% score by successfully executing a full churn-prevention sequence from start to finish without human intervention.
### A Shift in Anthropic’s Product Lineup
Anthropic’s product lineup consists of four tiers:
1. **Haiku**: Fast and cost-effective.
2. **Sonnet**: Mid-range performance.
3. **Opus**: The heavy-duty workhorse.
4. **Mythos**: The top tier, including Claude Fable 5 (public) and Claude Mythos 5 (restricted access for vetted researchers).
Fable 5 had a turbulent launch—pulled globally three days after its release due to a jailbreak vulnerability, then returning briefly under a credits-only model before being sidelined. Opus 5 now fills this gap, becoming the default choice for everyday users.
### Beyond Benchmarks: Real-World Impact
Developer platform Lovable reported that Opus 5 delivers improvements beyond scores: “It isn’t just better on our hardest agentic coding tasks, up 22% over Opus 4.7, it’s steadier, with far less variance run to run.”
Ultima Genomics, a DNA sequencing company, described the model as behaving “more like a careful scientist than any model we’ve run,” highlighting its ability to select appropriate statistical tests, cross-check results, and maintain focus through complex multi-step analyses.
### Availability and Pricing
Opus 5 is available via API at $5 per million input tokens and $25 per million output tokens. A faster mode is also available at twice the base price. The model is also accessible through subscription plans, making it widely available to all users.
—
## FAQ
**Q: How much does Claude Opus 5 cost?**
A: Claude Opus 5 costs $5 per million input tokens, identical to Opus 4.8 and half the price of Fable 5.
**Q: How does Opus 5 compare to GPT-5.6 Sol and Fable 5?**
A: Opus 5 outperforms both on most benchmarks. On Frontier-Bench v0.1, it scored 43.3% versus 34.4% for GPT-5.6 Sol and 33.7% for Fable 5. On ARC-AGI-3, it scored 30.2% compared to GPT-5.6 Sol’s 7.8%.
**Q: Is Claude Opus 5 available to all users?**
A: Yes. Opus 5 is available via API and through subscription plans on Claude Max and Claude Pro.
**Q: What happened to Claude Fable 5?**
A: Fable 5 was pulled shortly after launch due to a security vulnerability, returned briefly under a credits-only model, and has now been fully replaced by Opus 5 as the default flagship model.
—
## Conclusion
Claude Opus 5 represents a major step forward for Anthropic, combining frontier-level performance with cost efficiency and wide accessibility. By outperforming leading rivals on critical benchmarks while undercutting them on price, Opus 5 addresses the needs of both enterprise users and developers. With its robust performance across coding, problem-solving, and real-world workflows, Opus 5 is positioned to become the new standard in the AI model landscape.
