curatedai.net
Directory Categories Radar Models Leaderboard Guides Prompts Skills Briefing Newsletter
Light Dark
← Back
← All Claude models
LLMS • CURATED • UPDATED AUG 4, 2026

Claude Opus 5

Anthropic's frontier model, currently first on the Artificial Analysis Intelligence Index

Claude Opus 5 is Anthropic's flagship model, released 24 July 2026 with a 1M-token context window and five selectable effort levels (low, medium, high, xhigh, max). Effort is the main control: output token spend runs roughly 8x from low to max, and Artificial Analysis measures a 407-Elo spread in task quality across that range, so the same model behaves like several different price and capability tiers. API pricing is $5 per million input tokens and $25 per million output, with cache writes at $6.25 and cache hits at $0.50. It leads the Intelligence Index at 60.7 and tops the Coding Agent Index, scoring 89 percent on Terminal-Bench v2.1 at max effort and 53 percent on Humanity's Last Exam.

QUICK FACTS

Pricing Freemium
Platforms web, api, desktop
API Yes
Open Source No
Modalities LLMs, IDEs & Coding Tools, Multimodal Reasoning
Best For Best Frontier Model Overall
Date Added 2026-08-04

WHY WE PICKED IT

It is the current number one on the independent Artificial Analysis Intelligence Index, and it got there while costing less per task than the model it displaced: $2.03 average per index task against Fable 5's $2.75. The effort dial is the reason to pick it over a fixed-tier model, because one integration covers cheap high-volume calls and expensive long-horizon agent runs.

STRENGTHS

LIMITATIONS

GETTING STARTED

Use it free in the Claude apps, or via the API at model id claude-opus-5. Start at medium effort and only move up if the task actually fails, because max effort can spend roughly 8x the output tokens of low for work that did not need it. If you send the same system prompt or codebase repeatedly, turn on prompt caching first: cache hits cost $0.50 per million against $5 uncached.

BENCHMARKS

Artificial Analysis Intelligence Index · llm
60.7
Rank #1 · 2026-08-04

QUICK TIPS

1 Set effort per call rather than globally, cheap calls do not need max
2 Cache long system prompts and repeated context before optimising anything else
3 Benchmark cost per completed task, not cost per token, when comparing to Fable 5
4 Reach for max effort on terminal and multi-file refactor work, where the gap is largest
5 Measure on your own tasks, index rankings move within a few points

OFFICIAL LINKS

Website↗ API Docs↗ Artificial Analysis↗

SIMILAR TOOLS

Google Antigravity 2.0 Claude Fable 5 GPT-6 Astra NotebookLM GPT-5.6 Sol

USE CASE EXAMPLES

Cost-Tiered Agent Pipeline

Run one model across a pipeline where steps differ wildly in difficulty.

STEPS:
  1. Classify each step as routine or hard
  2. Send routine steps at low or medium effort
  3. Send planning and debugging steps at xhigh or max
  4. Cache the shared system prompt across every step
  5. Log tokens and cost per step
  6. Move steps down a tier until quality drops, then stop

Whole-Repository Review

Load a large codebase into the 1M-token window and review it in one pass.

STEPS:
  1. Concatenate the repository with file path headers
  2. Cache that context so follow-up questions are cheap
  3. Ask for an architecture summary first
  4. Follow with targeted questions on specific modules
  5. Request a prioritised defect list with file and line references
  6. Verify each finding against the actual file before acting

PRICING

Freemium Free tier available

Free tier includes limited features. Paid plans unlock full access, higher usage limits, and commercial usage rights.

📚

FEATURED IN GUIDES

CLAUDE.md In Practice: What Actually Earns A Place In The File

The most-starred CLAUDE.md on GitHub has four rules, not twelve, and the error rates everyone quotes...

Open Models: What They Cannot Do

The benchmark gap has largely closed: open models now score within a point of the closed frontier on...

Hardware For Local AI Models: What The Memory Shortage Changed

Capacity decides whether a model runs. Bandwidth decides how fast it talks, and it is the number mos...

How To Run AI Models Locally: Hardware, Quantisation And Engines

Running a frontier open model locally is a memory problem before it is a speed problem, and a bandwi...

Uncensored AI Models: What Abliteration Actually Does To Them

Abliteration removes a model's ability to refuse by deleting one direction from its activations. It ...

EXPLORE ALTERNATIVES

View Claude Opus 5 Alternatives (2026) →

Compare Claude Opus 5 with 5+ similar llms AI tools.

⚙️

KEY FEATURES

Capability

  • →Artificial Analysis Index 60.7 (#1 ranked)
  • →1M token context window
  • →Five effort levels (low-max) for cost control

Performance

  • →Strong reasoning tasks
  • →89% on Terminal-Bench v2.1
  • →Agentic autonomy excellence

Pricing

  • →$5/M input tokens
  • →$25/M output tokens
  • →Cache hits at $0.50/M (90% savings)
❓

TOOL-SPECIFIC FAQ

Q

What is 'effort' and why does it matter?

A

Effort (low, medium, high, xhigh, max) trades latency for quality. Low is fast and cheap, max is slowest and best. Same model weights, different compute budgets. Use low for simple tasks, high for complex reasoning.

Q

Is Claude Opus 5 worth the cost compared to Sonnet 5?

A

Yes for: reasoning tasks, coding, document analysis, complex instruction following. No for: simple queries (use Sonnet). Cost is 5-25x higher depending on effort level.

Q

How good is the 1M token context?

A

Excellent and actually usable. Can process entire codebases, long books, multiple documents. Effective context truly works. Output token cost scales with input.

Q

What is prompt caching and when should I use it?

A

Cache writes cost $6.25/M, reads cost $0.50/M (90% savings). Break even after 12+ identical prompts. Use for system instructions, customer docs, knowledge bases.

Q

Can I use Claude Opus 5 for production agents?

A

Yes - it's top-ranked for agentic autonomy. Higher cost but lower error rate. Many enterprises use it for critical autonomous workflows.

🏷️

FEATURED ON CURATEDAI

Work on Claude Opus 5? You're hand-reviewed in our directory. Add this badge to your site — it links back to this profile.

Featured on CuratedAI