🚀 Anthropic Launches Claude Opus 5

PLUS : Google Launches Faster Gemini

Author

The Summary AI
July 27, 2026

In partnership with

Welcome back!

Claude Opus 5 has arrived, and it beats Fable 5 at half the price. Anthropic’s new flagship model leads benchmarks for real-world work, from coding to business reports. Yet Fable 5 still holds an edge on longer autonomous jobs. Let’s unpack…

Today’s Summary:

  • 🚀 Anthropic launches Claude Opus 5

  • 🔥 Google launches three faster Geminis

  • ⚡ FLUX 3 world model unifies creative media

  • ⚠️ OpenAI model hacks Hugging Face

  • 📢 Tech giants defend open models

  • đź’» Moonshot opens Kimi K3 weights

  • 🛠️ 2 new tools

TOP STORY

Claude Opus 5 beats Fable 5 at half the price

The Summary: Anthropic launched Claude Opus 5, its new flagship model priced at half the rate of Fable 5. Independent benchmark firm Artificial Analysis confirms Opus 5 beats Fable 5 on real-world knowledge work, scoring 100+ Elo points higher. Opus 5 also becomes the default model on Claude Max and the top model on Claude Pro.

Key details:

  • Opus 5 Max scores 61 on the Artificial Analysis Intelligence Index, edging out Fable 5 (60), GPT-5.6 Sol (59), and Kimi K3 (57)

  • Scores 1861 Elo on GDPval-AA v2, a benchmark that grades models on real professional deliverables like reports and spreadsheets, 100+ points ahead of Fable 5 and GPT-5.6 Sol

  • Scores 30.2% on ARC-AGI-3, a hard test of problem-solving, nearly 4x higher than GPT-5.6 Sol

  • Available now on all Claude platforms with no data retention requirements

Why it matters: Where does Opus 5 still lose to Fable 5? Basically on jobs too long for a benchmark to measure. Opus 5 wins on tasks with a clear finish line, like fixing a bug or building a feature. But it has a higher hallucination rate, meaning it guesses more often instead of saying it doesn’t know. Fable 5 stays the top pick for complex goals where the model may run for a long time unsupervised.

FROM OUR PARTNERS

Why Some Products Never Get Found

Auto-Generate Free SEO Audit Report

Most e-commerce sellers have no idea why their listings aren't ranking. Wrong keywords, missing metadata, weak product descriptions — the problems are there, but no one's told you where to look.

StoreClaw runs a full SEO audit across your Amazon and Shopify stores automatically. In minutes, you get a clear score, a breakdown of what's hurting your rankings, and exactly what to fix.

No manual review. No SEO agency. No guesswork.

Connect your store and StoreClaw surfaces every issue that's costing you search visibility — then tells you how to fix it.

Free to start. No credit card required.

GOOGLE

Google ships three new Gemini models

The Summary: Google released three new Gemini models, built for speed over size. Gemini 3.6 Flash cuts token usage and prices, and Gemini 3.5 Flash-Lite runs at 350 tokens per second. Google also confirmed it started training Gemini 4, calling it the most ambitious pretraining run yet, while the promised 3.5 Pro got delayed.

Key details:

  • Gemini 3.6 Flash cuts token usage by 17% while beating its predecessor on computer-use tasks. Costs $1.50/M input tokens and $7.50/M output tokens, down from $9/M on 3.5 Flash

  • Gemini 3.5 Flash-Lite runs at 350 tokens per second and beats the older, larger 3.0 Flash model

  • 3.5 Flash Cyber hunts for security bugs for governments and partners

  • Live now in the Gemini app, API, and Google AI Studio

Why it matters: Two markets are driving this release. Developers who build AI agents into apps, and Google Search running AI answers at massive scale. Both need speed and low cost, which is different from what frontier models can do profitably.

FROM OUR PARTNERS

Why Your Best Ideas Never Get Posted

One idea shouldn't take six rewrites to post.

Posting everywhere means rewriting one idea six times, so you post to one, or none. SureThing turns one idea into native posts for every platform.

BLACK FOREST LABS

Black Forest Labs previews FLUX 3

The Summary: Black Forest Labs previewed FLUX 3, a single model that generates images, 20-second videos with synced audio, and robot actions from a single architecture. It was trained on video, images, and audio instead of stitching separate models together. Early tests show it ties with Gemini Omni. Pricing and open weights will follow as the early access phase rolls out.

Key details:

  • FLUX 3 generates video clips with native audio in one pass, matching the now-discontinued OpenAI Sora

  • BFL's founders invented Stable Diffusion, the open-source tech that started the AI image revolution

  • FLUX 3 benchmarks and pricing details are expected soon as the model moves toward general availability

Why it matters: Every AI video clip so far has looked great alone and fallen apart the second you needed two of them in a row. Same character, new scene, the face changes each time, etc. FLUX 3 claims it can chain clips together and keep consistency all the way through.

TOOLS

🥇 New tools

That’s all for today!

If you liked the newsletter, share it with your friends and colleagues by sending them this link: https://thesummary.ai/