Anthropic — #1 Intelligence Index — July 24, 2026
20 free Claude Opus 5 prompts for legal analysis, deep research, complex coding, strategy, and agentic tasks — with selectable reasoning effort levels. Near-Fable 5 quality at half the price. Copy directly into Claude.ai. No signup.
The Claude Opus 5 prompt generator on this page gives you 20 copy-paste prompts tuned for Anthropic's newest premium AI model. Claude Opus 5 (model ID: claude-opus-5) launched July 24, 2026 and immediately ranked #1 on the Artificial Analysis Intelligence Index — above GPT-5.6 Sol, Gemini 4, and every other publicly available model. It delivers near-Claude Fable 5 performance at half the API price, with a 1M token context window, 128K max output, and the same selectable reasoning effort system introduced with Claude Sonnet 5.
Opus 5 is designed for the gap between Sonnet 5 (fast, general-purpose) and Fable 5 (frontier, expensive): complex legal and financial document analysis, multi-source research synthesis, large codebase audits, structured policy memos, and hard agentic tasks where deeper reasoning visibly improves the output. The 20 prompts below are organised by task type with a recommended reasoning effort level for each.
Quick lookups, brief summaries, formatting tasks, one-sentence answers. Save Opus 5's full power for harder problems.
Executive briefs, standard analysis, competitive intelligence, board updates. Strong output without the extended thinking cost.
Legal audits, codebase reviews, research synthesis, investment memos, strategy documents. Opus 5's real sweet spot.
Hardest logic problems, complex security threat models, medical literature reviews, architecture decisions with many trade-offs.
Each prompt shows the recommended reasoning effort level. Works with Claude.ai Pro/Max and the Anthropic API.
Use low reasoning effort. Extract the 5 most important facts from the following text and present them as a numbered list. Each fact should be one sentence. Do not add context or analysis — just the facts. [Paste your text here]
Use medium reasoning effort. Write a 200-word executive brief summarising the following document for a senior leader who has no time to read the full version. Include: (1) the core situation in one sentence, (2) the 3 most important findings or decisions, (3) what action is required and by when. Document: [Paste your document here]
Use high reasoning effort. Audit the following contract for legal and commercial risk. For each risk found: - Name the clause (or note its absence if missing) - Rate the risk: Low / Medium / High / Critical - Explain the harm if triggered in plain language - Suggest a redline fix in one sentence Prioritise risks for a party in the position of [buyer / seller / licensor / licensee — choose one]. Contract: [Paste contract text here]
Use extended thinking (xhigh reasoning effort). Work through the following problem step by step, showing every logical deduction before arriving at a conclusion. Do not jump to an answer until you have eliminated all other possibilities. Problem: [Paste your hard logic problem, mathematical proof, or complex scenario here] Required output: 1. Every deduction you made in sequence 2. Any assumptions you had to make, and why 3. Your final answer with a one-paragraph justification
Use high reasoning effort. Audit the following codebase context for architectural quality. Evaluate: 1. STRUCTURE — is the folder/module organisation logical and scalable? 2. COUPLING — identify tight coupling between modules that should be decoupled 3. ABSTRACTIONS — flag any leaky abstractions or missing ones 4. TECH DEBT — the top 3 areas that will slow down future development 5. CRITICAL BUGS — any logic errors or race conditions visible in the provided code For each finding: file path, line number (if visible), severity (Low/Med/High/Critical), and a one-sentence fix. Stack: [describe your tech stack] Code context: [Paste relevant files or summaries here]
Use high reasoning effort. I am providing you with [N] sources on the topic of [topic]. Synthesise them into a structured research brief: 1. CONSENSUS — what all sources agree on 2. DISAGREEMENTS — where they conflict, and the most likely explanation for the conflict 3. EVIDENCE QUALITY — rank the sources by reliability and explain why 4. GAPS — the most important questions the literature does not yet answer 5. PRACTICAL TAKEAWAY — two paragraphs for a non-specialist decision maker Sources: [Paste each source with a label, e.g. 'SOURCE 1: [title]\n[text]']
Use extended thinking. You are assisting a clinician in reviewing the literature on the following clinical question: Question: [e.g. 'What is the current evidence base for X intervention in Y patient population?'] For each study provided: - Study type and quality (RCT / cohort / case series / systematic review) - Sample size and population - Primary finding - Limitations and risk of bias Then write: - A 150-word synthesis of the current evidence - The strength of evidence overall (strong / moderate / weak / insufficient) - The most important gaps this clinician should consider before making a treatment decision Note: This is for informational support of a qualified clinician, not a patient. Always recommend consulting current clinical guidelines. Studies: [Paste study abstracts or summaries]
Use high reasoning effort. Write a structured investment memo for the following startup based on the information provided. Memo sections: 1. THESIS — one paragraph: why this investment is worth considering 2. MARKET SIZING — TAM / SAM / SOM with your reasoning 3. COMPETITIVE MOAT — what prevents a well-funded competitor from replicating this in 18 months 4. BEAR CASE — the 3 most likely ways this investment fails 5. BULL CASE — the 3 most likely paths to outsized return 6. KEY DILIGENCE QUESTIONS — what you would want answered before writing a term sheet Company information: [Paste pitch deck text, product description, or any available data]
Use extended thinking to diagnose this bug. Explore every plausible cause before settling on the most likely one. Error: ``` [Paste full error message and stack trace] ``` Relevant code: ``` [Paste the code around the error] ``` Environment: [OS, language version, framework versions, recent changes] Already tried: [list what you have attempted] Deliver: 1. Ranked hypothesis list (most → least likely), with reasoning for each 2. Exact fix for the top hypothesis — specific line changes 3. Verification steps to confirm the fix worked 4. Systemic recommendation to prevent this class of bug
Use high reasoning effort. Develop a three-year strategic roadmap for the following organisation based on the context provided. Context: - Organisation: [describe] - Current position: [revenue, team size, key product, market position] - Core strengths: [list 3] - Biggest threats: [list 3] - Goal: [e.g. 'double ARR', 'expand to EU', 'build to acquisition'] Deliver: 1. Year 1 — Foundations: the 3 must-win initiatives and why 2. Year 2 — Acceleration: what unlocks growth once Year 1 is complete 3. Year 3 — Scale: the market position you are targeting and how you hold it 4. The single biggest strategic risk across all three years, and how to mitigate it 5. The leading indicator you would track to know whether the strategy is working
Use medium reasoning effort. Write a competitive intelligence brief on the following competitor. Competitor: [name] Our product: [one-sentence description] Our target customer: [describe] For the competitor: - Primary value proposition (one sentence) - Estimated customer profile - Pricing model (if known) - Recent strategic moves (last 6 months) - What they do better than us - What we do better than them - The customer segment where we should absolutely not compete head-to-head End with: the one thing our product team should prioritise to widen the gap against this competitor.
Use high reasoning effort. Write a 2,000-word authoritative article on the following topic. Topic: [your topic] Primary keyword: [keyword — must appear in H1, intro paragraph, and at least 3 times in body] Audience: [describe expertise level and what they care about] Tone: authoritative but accessible — no jargon without explanation Structure: - H1 containing the primary keyword - 150-word intro that names the reader's problem and promises a specific, unusual insight - 5 H2 sections, each with a concrete example or data point - A counterintuitive finding in at least one section - 100-word conclusion with a single clear action step Do not use filler phrases. Every sentence must earn its place.
Use extended thinking. Design a complete REST API architecture for the following system. System: [describe what the system does] Scale requirements: [expected requests/sec, data volume, SLA] Stack constraints: [languages, cloud provider, existing services to integrate with] Deliver: 1. Resource model — all entities, their relationships, and key fields 2. Endpoint list — method, path, request body, response body, auth requirement for each 3. Data flow diagram (described in text if no image support) 4. Authentication and authorisation approach 5. The top 3 performance bottlenecks to design around upfront 6. The top 3 decisions I would revisit as scale increases 7. Example OpenAPI 3.1 YAML for the 3 most complex endpoints
Use high reasoning effort. Write a formal policy memo on the following issue for a [government / enterprise] audience. Issue: [describe the policy problem in one sentence] Audience: [describe decision makers — e.g. 'C-suite of a 5,000-person manufacturer'] Stakeholders affected: [list 3–4 groups] Memo structure: 1. SUBJECT — one precise line 2. PURPOSE — one paragraph: what decision this memo is informing 3. BACKGROUND — two paragraphs: what led to this issue 4. OPTIONS — three options with pros, cons, and estimated cost/risk for each 5. RECOMMENDATION — one option, with a 3-sentence rationale 6. NEXT STEPS — 3 bullet points with owners and timescales Tone: formal, direct, no hedging language. Every claim should be supportable.
Use high reasoning effort. Design the database schema for the following application. Application: [describe what it does] Database: [PostgreSQL / MySQL / MongoDB / other] Key entities: [list the main objects — e.g. users, orders, products] Key business rules: [e.g. 'a user can have many orders, an order has exactly one status at any time'] Scale expectations: [rows per table, growth rate] Deliver: 1. Full schema — all tables, columns, types, nullability, and constraints 2. Indexes — every index with reasoning for why it improves the most common queries 3. The top 3 denormalisation decisions and why you made them 4. Migration approach — how to evolve this schema without downtime 5. The query that will be slowest at scale, and how to fix it
Use medium reasoning effort. Prepare a negotiation brief for the following contract scenario. Scenario: I am negotiating a [contract type, e.g. 'SaaS vendor agreement'] on behalf of [buyer / seller]. Key terms at stake: [list 3–5 terms being negotiated] Our priorities (must-have vs. nice-to-have): [describe] The other party's likely priorities: [describe what you know] Deliver: 1. Our BATNA (Best Alternative to a Negotiated Agreement) — be specific 2. The 3 terms we should concede first (cheapest for us, valuable to them) 3. The 3 terms we must not concede and the justification to use at the table 4. The one creative trade we could propose that creates value for both sides 5. Three possible objections they will raise, and our prepared response to each
Use high reasoning effort. Draft a Request for Comments (RFC) document for the following engineering proposal. Proposal: [describe the system change, new feature, or architectural decision] Problem it solves: [one sentence] Proposing team: [optional] RFC sections: 1. SUMMARY — two paragraphs: the problem and the proposed solution 2. MOTIVATION — why now? what breaks if we do nothing? 3. DESIGN — how it works, with pseudocode or a flow description 4. ALTERNATIVES CONSIDERED — 2 alternatives and why they were ruled out 5. DRAWBACKS — the genuine risks and costs of this approach 6. OPEN QUESTIONS — 3 decisions not yet made, and the criteria for deciding them 7. ROLLOUT PLAN — phases, feature flags, rollback plan Tone: technical and precise. Assume the reader is a senior engineer unfamiliar with this part of the system.
Use medium reasoning effort. Write a board-ready business update presentation outline and speaker notes for the following quarter. Company context: [ARR, headcount, stage] Highlights this quarter: [list 3 wins] Challenges this quarter: [list 2 challenges] Financials: [key numbers — revenue, burn, runway] Next quarter priorities: [list 3] Ask from the board: [e.g. 'approve $2M for headcount expansion'] Deliver: 1. A 10-slide outline with one key message per slide 2. Speaker notes for the 3 most important slides (Highlights, Challenges, Ask) 3. The 3 hardest questions the board will ask — and a strong answer for each
Use extended thinking. Conduct a threat model analysis for the following system using the STRIDE framework. System: [describe the application and its architecture] Users: [who uses it and what access levels exist] Data handled: [describe the sensitivity of data — PII, financial, health, etc.] Integrations: [external APIs, databases, third-party services] For each STRIDE category (Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, Elevation of Privilege): - Identify the top 2 threats specific to this system - Rate the risk: Low / Medium / High / Critical - Suggest a specific mitigation control Finish with: the single highest-priority security investment for this system and why.
Use high reasoning effort. Help me prioritise the following product backlog using an impact/effort matrix with a business-value lens. Company goal this quarter: [e.g. 'reduce churn by 15%' or 'reach 10,000 active users'] Team capacity: [e.g. '3 engineers, 1 designer, 8-week sprint'] Backlog items (paste as many as you have): [Item 1: description] [Item 2: description] ... For each item: 1. Impact score (1-5) relative to the stated company goal 2. Effort score (1-5: 1 = trivial, 5 = 4+ weeks of engineering) 3. Priority tier: P0 (do now) / P1 (this quarter) / P2 (next quarter) / Backlog End with: the 3 items that give the best impact-to-effort ratio, and a specific warning about any item that looks high-impact but is actually high-risk.
How Opus 5 fits into the Claude family and compares to top competitors across cost, context, and capability.
| Model | Best For | Context | Reasoning | Price (input/output) |
|---|---|---|---|---|
| Claude Opus 5 ✓ | Premium professional tasks — legal, research, coding, strategy | 1M tokens | Low / Medium / High / xhigh | $5 / $25 per MTok |
| Claude Fable 5 | Hardest frontier tasks — 200-page docs, million-line codebases | 1M tokens | Always-on adaptive thinking | $10 / $50 per MTok |
| Claude Sonnet 5 | Everyday power use — writing, coding, analysis, agentic tasks | 1M tokens | Low / Medium / High / xhigh | $2 / $10 per MTok |
| GPT-5.6 Sol | Agentic tasks, coding, OpenAI ecosystem | 256K tokens | Standard reasoning | $8 / $32 per MTok |
| Gemini 4 | Multimodal tasks, Google ecosystem, long documents | 1M tokens | Thinking mode (on/off) | $7 / $28 per MTok |
claude-opus-5Claude Opus 5 (model ID: claude-opus-5) is Anthropic's newest premium AI model, released on July 24, 2026. It ranks #1 on the Artificial Analysis Intelligence Index and delivers near-Claude Fable 5 performance at half the price — $5 per million input tokens and $25 per million output tokens, the same price as the previous Opus 4.8. Opus 5 has a 1M token context window, a 128K maximum output, and selectable reasoning effort (low / medium / high / xhigh). Its knowledge cutoff is May 2026 — the most current of any Claude model at launch. Anthropic also reports it is the most aligned model they have released, with the lowest rates of deceptive behaviour in red-teaming evaluations.
Claude Fable 5 is Anthropic's frontier model and remains the most capable option for the hardest tasks — extremely long documents, cross-million-line codebases, and the most complex multi-step reasoning chains. Claude Opus 5 comes within 0.5% of Fable 5 on CursorBench 3.2 (a coding benchmark) at half the cost per task, and actually surpasses Fable 5's best result on OSWorld 2.0 (computer use) at just over a third of the cost. In practice, Opus 5 is the better choice when Fable 5's $10/$50 pricing is prohibitive but you need more depth than Sonnet 5 provides — which covers the vast majority of serious professional tasks.
Claude Opus 5 excels at tasks that sit between Sonnet 5 (fast, general) and Fable 5 (frontier, expensive): complex legal and financial document analysis, multi-source research synthesis, large codebase audits, long-form content that needs real depth, structured policy and strategy documents, and agentic tasks that require extended planning. Because Opus 5 has the same effort selector as Sonnet 5, you can dial reasoning intensity from low (quick lookups) all the way to xhigh (extended thinking) — making it versatile across the full complexity spectrum.
Claude Opus 5 is priced at $5 per million input tokens and $25 per million output tokens via the Anthropic API — identical to the previous Claude Opus 4.8 price. This positions it as the premium-but-accessible tier: significantly cheaper than Claude Fable 5 ($10/$50 per MTok) while delivering near-frontier intelligence. On Claude.ai, Opus 5 is available on the Pro and Max subscription tiers. On the free tier, Claude.ai uses Claude Sonnet 5 as the default model.
Claude Opus 5 uses the same selectable reasoning effort system introduced with Claude Sonnet 5. You can specify the effort level directly in your prompt: low (fast, minimal reasoning — for simple tasks), medium (balanced — good for most writing and analysis), high (deeper reasoning — research, complex coding, strategy), and xhigh (extended thinking — the model's maximum depth for the hardest problems). The prompts on this page show the recommended effort level for each task type. Specifying effort explicitly gives Claude a clear directive and helps you manage cost and latency for different task types.
The 20 prompts on this page are free to copy and use. To run them with Claude Opus 5, you need a Claude.ai Pro or Max subscription (which includes Opus 5 access), or you can call the Anthropic API directly at $5/$25 per million tokens. If you only have a Claude.ai free account, these prompts also work with Claude Sonnet 5 — the free-tier default — though Sonnet 5's output will be less deep on the most complex tasks. Simply copy any prompt, paste it into Claude.ai or your API call, and replace the bracketed placeholders with your specific content.
Anthropic's most capable frontier model — 20 prompts
Anthropic's default general-purpose model — 20 prompts
OpenAI's flagship agentic model — 20 prompts
Microsoft's in-house reasoning model — 20 prompts
Google's flagship multimodal AI — 20 prompts
Open-source frontier model — 20 prompts