Compare Start here

AnthropicLanguage modelSuperseded

Claude Opus 4.5

Cut Opus pricing by two thirds and took the SWE-bench crown in late 2025.

Compare head-to-head Race it in 3D

Overall Intelligence

#30

of 56
12.8 pts behind #1

0

races led

135

weeks at #1

75.9

score

Released: Nov 24, 2025; 11 mo ago
Context: 200K; 64K max output
Price: $5 / $25; per 1M in/out · $10.00 blended
Speed: 58 tok/s; 5s to first token
Modalities: text · image; outputs text
Weights: Closed; API and apps

The 30-second version

The late-2025 Opus that made Claude the default for coding agents, and held the agentic coding lead for most of the winter.

Key takeaways

1Long reign at the top of agentic coding.
2Big price cut versus Opus 4.1.
3Popularised long-running Claude Code sessions.

Where it ranks

Current snapshot · illustrative scores · click any race to watch it

Opus 4.5Best on each axis

Overall

30Overall+12.8 pts

Coding

25Coding+12.7 pts 23Frontend & UI+68 Elo 24Backend & APIs+10.7 pts 30Distributed Systems+15.6 pts 26Full-Stack Apps+12.9 pts 26Debugging & Repair+14.0 pts 25Agentic Coding+13.8 pts 28Code Review & Security+12.4 pts 27Data Science & SQL+11.1 pts 19Mobile Apps+8.9 pts 24Terminal & DevOps+14.7 pts 29Competitive Programming+421

Hard Reasoning

30Hard Reasoning+14.9 pts 34Frontier Questions+27.4 pts 31Abstract & Novel Reasoning+22.1 pts 35Logic & Puzzles+13.5 pts 27Planning & Strategy+12.5 pts

Math

34Math+16.1 pts 36Competition Math+13.7 pts 34Research Math+20.3 pts 30Applied & Quantitative+12.2 pts

Science

29Science+13.9 pts 30Graduate Science QA+9.2 pts 26Biology & Medicine+11.3 pts 32Physics & Chemistry+12.9 pts 27Research Assistance+11.9 pts

Writing & Language

26Writing & Language+11.9 pts 20Creative Writing+68 Elo 21Technical Writing+9.3 pts 26Editing & Style+9.2 pts 27Multilingual+8.3 pts 32Instruction Following+10.2 pts 28Conversation & EQ+75 Elo

Multimodal

34Multimodal+26.7 pts 30Image Understanding+13.7 pts 30Documents & Charts+14.3 pts 33Video Understanding+27.7 pts 37Speech & Audio+38.6 pts

Tool Use & Agents

24Tool Use & Agents+12.9 pts 25Tool Use+9.8 pts 20Computer Use+16.8 pts 29Web Research+17.1 pts 28Long-Horizon+23.2 hrs

Long Context

31Long Context+16.7 pts 341M Recall+14.7 pts 28Long Reasoning+13.8 pts

Speed & Cost

48Speed & Cost+51.6 pts 46Output Speed+463 tok/s 35Latency+4.42s 50Cost-Efficiency+61.1 pts

Trust & Factuality

27Trust & Factuality+13.0 pts 26Factuality+10.1 pts 31Honesty & Calibration+11.4 pts

Overall Intelligence over time

#1 frontierOpus 4.5GLM-5.3Large 4

Update history

The distilled version first; open any release for the deeper breakdown.

The Claude Opus family

Opus 4May 2025 Opus 4.1August 2025 Opus 4.5November 2025 Opus 4.6February 2026 Opus 4.8May 2026 Opus 5July 2026 Opus 5.5September 2026

60-second briefing

Opus 4.5, explained

AI video summary · placeholder render from the auto-written script

From Anthropic's own guide, distilled

Prompt engineering and Claude best practices

Be explicit about the outcome you want; newer Claude models follow instructions literally rather than guessing at hidden intent.
Explain why a rule exists. Context about the goal lets the model generalise the rule to cases you did not list.
Wrap documents, examples and data in XML-style tags so instructions and material never blur together.
Put long documents at the top of the prompt and your question at the end; recall and grounding both improve.
For agentic tasks, state when to act versus when to ask, and ask for a short plan before large edits.

All official guides, distilled →

About Anthropic

Safety-focused lab building the Claude family of models.

Optimising for

Reliable long-horizon agentic work (especially coding and computer use), honest and calibrated answers, and strong writing, with safety research shipped alongside capability.

San Francisco, US · founded 2021

Closest rivals

Opus 4.5 is 30th in overall intelligence; leader is Claude Opus 5.5.