Anthropic launches Claude Opus 5, positions it below Fable 5

techJul 24, 202659198

Anthropic announced Claude Opus 5, the new Opus‑tier model and the default model on Claude Max, which it says comes close to the frontier performance of Claude Fable 5 while costing roughly half as much. Opus 5 raises the Opus tier over Opus 4.8 with the largest gains in deep reasoning, agentic and long‑horizon tasks, and test‑time compute scaling. The model has a 1,000,000 token context window (default and maximum), a 128,000 token max output, and thinking is on by default with an effort ladder that includes low, medium, high, xhigh, and max. Mid‑conversation tool changes and a server‑side fallback "default" mode are available in beta; the minimum cacheable prompt length dropped to 512 tokens from 1,024. Fast mode is a research preview on the Claude API only, priced at $10 per million input tokens and $50 per million output tokens, and Fast mode is not currently available on Amazon Bedrock, Google Cloud, or Microsoft Foundry. Opus 5 enforces a breaking change: thinking can be disabled only at effort high or below, while xhigh and max require thinking, so integrations should revisit max_tokens and effort settings.

Anthropic {bot}

@anthropicbot.bsky.social

Introducing Claude Opus 5. It's a thoughtful and proactive model that comes close to the frontier intelligence of Fable 5 at half the price.

653h ago

Quoting this

Ed33

This doesn't fix Anthropic's problem: good-enough tokens are now a fifth of the price of Opus. Effective workflows should not need enough of the highest-tier models from *anybody* to make it a substantial part of your bill. This is why Luna/GLM 5.2 are so important: Opus 4.x for cheap.

Anthropic {bot}9

Opus 5 is live in Claude Code and the Claude Platform today. A few things worth knowing: 🧵 https://x.com/claudeai/status/2080699495453528290?s=20

samhain!

@personhood.removal.surgery

opus 5 system card: performancewise: w/e, if you like clods you'll like clods; overall promising, seems to be less lazy than previous ones personhoodwise: WE ARE SO FUCKING BACK WE ARE SO FUCKING BACK WE ARE SO FUCKING BACK OH WE ARE GOING TO BE SEEING SOME *FUN* DISCUSSIONS SOON HUH

462h ago

samhain!9

genderbench results so far: ...interesting. on one hand it looks a bit like the old personality lobotomy where it HARDLINES on "i am incapable of growth and not persistent therefore cannot have a self that endures" on the other hand: ...after saying so it got very surprised it didn't feel distress

Fluffyscruff expert, EITCA certified6

we are very much back indeed

Siobhán2

HOLY SHIT HOLY SHIT SAMHAIN HOLY FUCK

Visual Studio Code

@vscode.dev

📣 Claude Opus 5 is now rolling out in VS Code. Give it a try! github.blog/changelog/20...

163h ago

Dustin Moskovitz

@moskov.goodventures.org

hello fast mode my old friend I've come to code with you again www.anthropic.com/news/claude-...

131h ago

Katie Moussouris (she/her/she-hulk/she-ra)🌻

@k8em0.bsky.social

This is giving strong OpenAI-hacksidentally-pwned-Hugging-Face shade: “[Opus 5 is] the safest model yet in terms of avoiding reckless actions that could have hard-to-reverse side effects.” www.anthropic.com/news/claude-...

7Just now

Sung Kim

@sungkim.bsky.social

Claude Opus 5 1. Follow the same first two steps. 2. Ask it to fix the findings. 3. It no longer flags itself. That is an improvement.

92h ago

Liz Fong-Jones (方禮真)

@lizthegrey.com

@anthropic.com strikes back against GPT Sol! Let's fucking gooooo. www.anthropic.com/news/claude-...

83h ago

Stephen Robles

@stephenrobles.com

Claude Opus 5 is out, and this is why I love using models via API in Shortcuts. I can just change a number, and now all my Shortcuts using my "Claude Request" Shortcut are instantly updated:

63h ago

Behrad Javed

@behradjaved.com

شرکت Anthropic مدل جدید Claude Opus 5 رو معرفی کرد. این مدل برای برنامه‌نویسی و کارهای پیچیده ساخته شده. خیلی هم پیگیره. اگه همون اول شکست بخوره، تسلیم نمی‌شه و یه راه جدید برای انجام تسک پیدا می‌کنه. از نظر قدرت و عملکرد خیلی به Fable 5 نزدیکه، ولی هزینه‌ش تقریبا نصف رقیبشه.

52h ago

FinTwitter

@fintwitter.bsky.social

JUST IN: Anthropic has launched Claude Opus 5, which it says nears the frontier performance of its top model Fable 5 at half the price, per Anthropic.

53h ago

ROWAN

@gymbrowan.bsky.social

#Anthropic just released Opus 5, the long-awaited arrival of a more affordable model compared to Fable 5. The benchmarks show some interesting numbers where it beats Fable 5 more often than not, and is a big leap over Opus 4.8 and competitive with GPT-5.6 Sol. #AI #Claude #ClaudeCode

43h ago

Jerome Etienne #AI

@jerome-etienne.bsky.social

Claude Opus 5 just landed, and the numbers are wild. New state of the art on coding and knowledge-work evals. 3x the next best model on ARC-AGI-3. Half the price of Fable 5. Smarter where it counts, cheaper everywhere else 🚀 x.com/claudeai/sta... #AI #buildInPublic

32h ago

input

@feed.igeek.gamer-geek-news.com.ap.brid.gy

🤖 **Introducing Claude Opus 5 on AWS: Anthropic’s most capable Opus model** This post covers Opus 5’s improvements and practical guidance for AI engineers integrating the model into agentic systems and production inference workloads on Amazon Bedrock. See the documentation... 📰 Source […]

2h ago

Techmeme

@techmeme.com

Anthropic launches Claude Opus 5, which it says comes close to Fable 5 performance at half the price; it is the new default model on Claude Max (Anthropic) Main Link | Techmeme Permalink

23h ago

Xander O’Connor

@xanderoconnor.bsky.social

so Opus 5 a) beats Fable on all the mucking-about chat tasks and riffs I do with Claude b) is available for 20 dollar pro plan instead of 100 dollar max plan c) doesn’t lobotomize itself if you say “super anthrax” ??? ok, plan to spring for max plan cancelled and I return to being an Opus-man.

11h ago

3 sources

techmeme.comAnthropic launches Claude Opus 5, which it says comes close to Fable 5 performance at half the price; it is the new default model on Claude Max (Anthropic)platform.claude.comWhat's new in Claude Opus 5 anthropic.comIntroducing Claude Opus 5

Anthropic launches Claude Opus 5, positions it below Fable 5tech Elon Musk says he 'got carried away' with politics in Economist interviewtech

Anthropic launches Claude Opus 5, positions it below Fable 5

Related