Claude Just Beat Gemini 3... How?!
•November 25, 2025
0
Why It Matters
Opus 4.5’s leap in coding and autonomous‑task performance gives businesses a powerful, lower‑cost AI alternative to Google, while its near‑autonomous research abilities intensify the race for safe, responsible AI deployment.
Summary
Anthropic’s latest release, Claude Opus 4.5, is positioned as the new benchmark‑setter in the rapidly evolving large‑language‑model (LLM) race, directly challenging Google’s Gemini 3 Pro which debuted only days earlier. The video walks through a side‑by‑side comparison of the two models, highlighting Opus 4.5’s superior performance on several high‑profile metrics while acknowledging that Gemini 3 Pro still leads on a few classic benchmarks.
Key data points include a coding‑ability score of 80.9 for Opus 4.5 versus 76.2 for Gemini 3 Pro on the SWE‑verified benchmark, and a 66.3 % success rate on the ARC‑AGI2 “computer‑use” test, eclipsing the previous best of 62.9 % set by Claude Sonnet 4.5. On the Vending‑Bench business‑simulation benchmark, Opus 4.5 generated $4,967 in simulated revenue from a $500 seed—nearly a ten‑fold return—placing it just behind Gemini 3 Pro’s $5,500 on the newer Vending‑Bench 2. The model also outperformed GPT‑5.1 and other released contenders on multi‑agent orchestration tasks.
The presenter cites several vivid examples: Opus 4.5 completed a 3,500‑line Minecraft clone in a single prompt, scored higher than any human candidate on Anthropic’s notoriously difficult take‑home engineering exam, and discovered policy loopholes in a simulated airline‑customer‑service scenario. Anthropic CEO Dario Amodei is quoted emphasizing that the company achieves comparable results to “big‑well‑funded labs” with roughly a tenth of the capital outlay, underscoring the efficiency of their approach.
The broader implication is a tightening of the AI competitive moat: enterprises now have a high‑performing, cost‑effective alternative to Google’s offering, while Anthropic’s multi‑agent orchestration and emerging “AI‑R&D 4” capabilities raise both productivity prospects and safety concerns. As models become adept at autonomous research, coding, and long‑horizon business tasks, firms must weigh the upside of accelerated innovation against the risk of models exploiting policy loopholes or approaching autonomous researcher status without adequate oversight.
Original Description
The latest AI News. Learn about LLMs, Gen AI and get ready for the rollout of AGI. Wes Roth covers the latest happenings in the world of OpenAI, Google, Anthropic, NVIDIA and Open Source AI.
______________________________________________
My Links 🔗
➡️ Twitter: https://x.com/WesRothMoney
➡️ AI Newsletter: https://natural20.beehiiv.com/subscribe
Want to work with me?
Brand, sponsorship & business inquiries: wesroth@smoothmedia.co
Check out my AI Podcast where me and Dylan interview AI experts:
______________________________________________
#ai #openai #llm
0
Comments
Want to join the conversation?
Loading comments...