AI Power, Half The Price.
Anthropic Launches Claude Opus 5, Offering Fable 5-Like AI Performance at Half the Price
Anonymous
Key Points
1
Opus 5's Frontier-Bench results show a 43.3% score, beating Fable 5's 33.7% and Opus 4.8's 18.7%, at a lower cost per task.
2
Anthropic describes Opus 5 within a broader lineup: Opus 5 as the daily driver, Fable 5 for days-long autonomous projects, Sonnet 5 for work at scale, and Haiku 4.5 for subagents and instant answers.
3
Opus 5's safeguards are less restrictive than Fable 5's because it is less capable of exploiting cyber vulnerabilities, per Anthropic's testing.
4
Leaks suggest Opus 5 could serve as a fallback for Fable 5 guardrails during testing, with evidence of 'Opus 5 High' and session handoffs from Fable 5 to Opus 5.
5
Anthropic's comments on open-weight models, including Moonshot's Kimi K3, indicate that how open-weight models perform on real-world projects remains to be seen.
Anthropic has launched Claude Opus 5, a new AI model that nearly matches its flagship Fable 5 at roughly half the price. According to MacRumors, Opus 5 scores 43.3% on Frontier-Bench, beating Fable 5's 33.7% and the older Opus 4.8's 18.7%, all while costing $5 per million input tokens and $25 per million output tokens.
Glitchwire reports that Opus 5 is now the default model on Claude Max and a strong option on Claude Pro. The launch signals a clear push by Anthropic to win mid-range and enterprise customers where cost per task matters most.
Opus 5 Beats Fable 5 on Key Benchmarks at Half the Cost
The headline number is striking. Opus 5's 43.3% Frontier-Bench score beats Fable 5's 33.7% by nearly ten points, Cryptobriefing reports. That gap is especially notable for coding and knowledge work, two areas Anthropic is directly targeting with this release.
NDTV Profit notes that Opus 5 handles complex tasks with less back-and-forth from users. It also checks its own work, reducing errors without requiring extra prompting. For businesses running high volumes of tasks, that efficiency adds up fast.
Where Fable 5 Still Wins — and Why Safeguards Differ
Fable 5 is not going away. Anthropic positions it for long, multi-day autonomous projects where its deeper capabilities matter. Opus 5, by contrast, is built for daily use. TechBuzz describes the full lineup as: Opus 5 for everyday tasks, Fable 5 for days-long work, Sonnet 5 for scale, and Haiku 4.5 for quick subagent answers.
Anthropic also says Opus 5's safety guardrails are less restrictive than Fable 5's. The reason, per MacRumors, is straightforward: Opus 5 is less capable of exploiting cyber vulnerabilities, so it needs fewer restrictions. U.S. government concerns over Fable 5's potential cyber capabilities had made those tighter limits necessary.
Leaks Point to 'Opus 5 High' and Fable 5 Handoffs
Before the official launch, leaks sparked wide attention. Evidence of a variant called 'Opus 5 High' surfaced online, along with signs of session handoffs from Fable 5 to Opus 5, Glitchwire reported. That raised speculation that Opus 5 could serve as a fallback when Fable 5's guardrails are triggered during limited testing.
Other rumors suggested Opus 5 could rival tools used in 3D generation and game design. Anthropic has not confirmed those claims. The leaks did, however, drive strong early interest from developers ahead of the formal announcement.
Anthropic Takes Aim at Enterprise Work With Aggressive Pricing
At $5 per million input tokens, Opus 5 undercuts Fable 5 significantly. Cryptobriefing notes the price matches Opus 4.8 while delivering far stronger performance. That combination makes Opus 5 a direct play for enterprise buyers who need capable AI at a predictable cost.
Anthropic also weighed in on open-weight rivals. The company said it remains to be seen how models like Moonshot's Kimi K3 perform on real-world projects, per Glitchwire. The message was clear: benchmarks are one thing, but production results are what matter for enterprise deals.
Sentiment
Mostly PositiveNEGATIVENEUTRALPOSITIVE
Publishers
33
Articles
108
Reach
141
A
AnonymousAuthor
