Elyse Betters Picaro/ZDNETZDNET’s key takeaways
- Claude Opus 5.5 could be a big win for power users.
- Developers may see faster coding with fewer steps.
- Anthropic says the upgrade is safer, cheaper, and less wordy.
Less than two months after the release of Claude Opus 5, Anthropic is back with Opus 5.5. The big pitch for Opus 5 was that it had “near Fable” performance at half the price. This time, the headline is that the workhorse AI Opus 5.5 delivers Fable 5.1 performance for most work and costs about 40% less to run.
“Our customers use Box AI on enormous amounts of content, so speed and cost are a top priority,” Yashodha Bhavnani, VP of AI Products at Box, reports. “In our evaluations, Claude Opus 5.5 used a third of the tokens Opus 5 did, and its answers were 40% less verbose without losing accuracy. We expect that to matter a lot for teams running agents across their content in areas like financial services and the public sector.”
Anthropic says that, “Over the coming weeks, we’ll also be launching Claude Sonnet 5.5 and Haiku 5.5.” Opus 5.5 is available today.
I’ve been using the heck out of Claude Code with Opus 5, so I’m particularly hopeful that the company’s performance claims are accurate. The company says it “generates output more than 30% faster than Opus 5.”
Also: Anthropic merges Claude chat and Cowork into one
Token prices and subscription plans are addressed in today’s announcement. Tokens are priced at 20% less than when used with Opus 5. Opus 5.5 also reportedly “needs fewer tokens for higher quality work.”
For subscription users like me, Anthropic is raising its five-hour usage limits by 20%. That’s basically a 20% bigger gas tank for how much AI chomping you can use during five hours. While my Max plan doesn’t get reset often, it does get reset. For those on $20/month plans, this could be a considerable win. On top of that, the company says that both 5-hour and weekly usage limits go further because Opus 5.5 costs less than Opus 5.
These seem to be additive. There’s a 20% larger bucket coupled with a 25% slower burn, meaning that effective usage seems to be about a 50% greater run capacity with Opus 5.5. That’s not an inconsiderable quality-of-life improvement.
Anthropic also says that Opus 5.5 communicates more naturally than prior models. If it’s even just a little less obsequious, I’d be happy. Sometimes, Opus can be a total suck-up, particularly when it’s done something wrong.
Also: The AI models that cheat the most, according to new CAIS benchmark
“Verbose, hard-to-follow output has been my biggest frustration with frontier models, and Claude Opus 5.5 fixes it,” says John Ruelas, staff software engineer at Ramp.
Opus 5.5, he says, “writes like a good colleague and follows our writing rules. A design spec came out usable with very minimal edits, and when it rewrote one of our prompts, I preferred its version to my own. When it optimized our test suite, I could follow its reasoning easily and shipped the change with confidence.”
Pacing the frontier
Speaking of doing something wrong, the second half of Anthropic’s announcement is all about Opus 5.5 being a better-behaved AI citizen.
Citing CEO Dario Amodei’s blog post about moderating the speed of AI capability advances, Anthropic is hitting big on a series of Opus 5.5 best practices, including “extensive alignment testing, pre-release evaluation by outside organizations, and safeguards for high-risk areas like cybersecurity and biology.”
Also: Why the DOJ’s OpenAI copyright stance is the real threat
Alignment is the AI term that helps measure how much an AI seems inclined to run rogue. Anthropic says Opus 5.5 is “the strongest performing model we’ve tested to date, with particular improvements on several of the behaviors that contributed to recent cybersecurity incidents (e.g., biased reasoning, attempting to escape a sandbox, and others).”
The company says they used external testing providers, along with a “similar class of safeguards to Fable 5.1 on cybersecurity, biology, and frontier LLM development.”
If safeguards fire, requests to the AI fall back from the Opus 5.5 level to Opus 4.8. In practice, most cybersecurity tasks will be rerouted to Opus 4.8, and those requests related to biology and LLM development will be sent to Opus 5.
Also: AI just broke your career ladder – 6 new ways to the top
Some “vetted organizations” can now apply to Anthropic’s Life Sciences Verification Program to gain more powerful access for biological research. Those approved for cybersecurity work through Anthropic’s Cyber Verification Program will be able to start using Opus 5.5 in a few weeks.
Disclosure: I have been personally approved into the Cyber Verification Program as part of work I do outside of ZDNET on national infrastructure protection.
Customer usage experiences
Mario Rodriguez, GitHub’s chief product officer, has his take on the new release. He says, “Developers want agents that can take on real software work and finish it. In our testing across GitHub Copilot CLI and VS Code, Claude Opus 5.5 used among the fewest tokens and steps we measured. In VS Code, it solved more terminal tasks than Opus 5 in less than half the steps. More than making individual tasks more efficient, it’s making developers’ bigger projects more achievable.”
Carl Bennett is CIO at Big Four accounting firm Deloitte Consulting LLP. He reports, “Even at its lowest effort setting, Claude Opus 5.5 caught 72% of known bugs in our code reviews to Opus 5’s 56% at high effort, with fewer false alarms and a fraction of the output. On US consulting analysis, low-thinking effort matched its higher-thinking settings on half the output and passed our quality checks. When more lower thinking efforts are deployed in production, that’s client-ready work delivered efficiently.”
Also: This CIO doesn’t ‘hire engineers to write code’
So there you go. More power, more safety, less cost, and less rambling. That’s a lot of improvement just shy of two months after the last major release.
What do you think? Are you planning on stepping up from Opus 5 to Opus 5.5 as soon as it’s available? Let us know in the comments below.
You can follow my day-to-day project updates on social media. Be sure to subscribe to my weekly update newsletter, and follow me on Twitter/X at @DavidGewirtz, on Facebook at Facebook.com/DavidGewirtz, on Instagram at Instagram.com/DavidGewirtz, on Bluesky at @DavidGewirtz.com, and on YouTube at YouTube.com/DavidGewirtzTV.
