Utk yg mo Bantu2 Keuangan saya..monggo ke Bank Central Asia BCA 5520166779 a.n. Andreas Tparlaungan Manurung (Indonesia) For those who would like to help support my finances ..please feel free to send it to Bank Central Asia (BCA) account number 5520166779 under the name Andreas Tparlaungan Manurung (Indonesia)

🧠 Mega unified prompt-engineering comparison

OpenAI · Anthropic · Google (Gemini) · Meta (Llama) · Mistral AI — side by side

📊 Prompt engineering: feature‑by‑feature

Prompt element	OpenAI	Anthropic	Google / Gemini	Meta (Llama ecosystem)	Mistral AI
Primary prompting style	Clear, direct, instruction-first	Modular, structured, section-based	Spec-like, structured-output oriented	Explicit instruction + pattern examples	Lean, natural-language, clarity-first
Role / persona	Useful but often optional	Often explicit as separate role block	Usually folded into instruction	Often improves consistency, commonly recommended	Helpful but usually lightweight
Context / background	“Context”, “background info”, supporting material	“Background data”, documents, prior conversation	Context often treated as required task input	Important for grounding, especially with longer prompts	Strongly benefits from concise relevant context
Objective / task	Clear task definition is central	Explicit task block	Instruction/task is the anchor	Explicit task request strongly preferred	Direct task phrasing works best
Constraints / rules	Requirements, limits, exclusions	Rules, guidelines, behavioral instructions	Strong emphasis on explicit constraints	Explicit restrictions help stabilize output	Constraints should be concrete and simple
Output format	Desired output format / response structure	Output formatting section	Strong emphasis on structured output / schema-like control	Formatting instructions help noticeably	Works best with explicitly requested structure
Tone / style	Explicit tone/style guidance encouraged	Often separated as tone context	Usually included inside instruction	Stylization often works well	Best kept concise and direct
Examples / few-shot	Strongly recommended when output pattern matters	Very common and often treated as separate prompt block	Common for format control and consistency	Particularly useful for style imitation and pattern matching	Useful, but usually simple examples are enough
Reasoning guidance	Step-by-step can help depending on task	Often explicitly encourages thinking steps	Often benefits from decomposed task instructions	Chain-style prompting often useful	Clear decomposition helps more than verbosity
Best prompt shape	Task → Context → Format → Style	Role → Background → Task → Rules → Output → Examples	Instruction → Context → Constraints → Structured Output	Instruction → Context → Constraints → Examples	Goal → Context → Clear Ask
Typical strength	Fast practical usability and broad general-purpose prompting	Complex multi-part instructions, organized workflows	High control, structured outputs, deterministic formatting	Strong pattern-following and adaptable style behavior	Efficient prompting with minimal overhead
Common failure mode if prompt is weak	May answer too broadly	May become generic if structure is vague	May under-deliver detail if scope is not explicit	Can drift stylistically if examples are weak	Can become overly terse if underspecified

📌 Practical shorthand per ecosystem

OpenAI GPT series

Best shorthand: Context + Objective + Constraints + Format + Tone

Anthropic Claude

Best shorthand: Role + Context + Task + Rules + Output + Examples

Google / Gemini

Best shorthand: Instruction + Context + Constraints + Structured Output

Meta (Llama)

Best shorthand: Instruction + Context + Constraints + Examples

Mistral AI

Best shorthand: Goal + Context + Clear Ask

⚡ Fastest practical reading

🧭 “At a glance” style affinities

• OpenAI → best for clear direct prompting
• Anthropic → best for modular prompt documents
• Google → best for structured-output specifications
• Meta → best for instruction + examples
• Mistral AI → best for compact efficient prompting

---

💡 Why this matters
A lot of this aligns with official prompting guidance from OpenAI and Anthropic, plus public prompt-engineering materials around Google. Each model family has unique sensitivities: OpenAI handles concise instructions beautifully, Anthropic thrives on structured sections, Gemini loves explicit schemas, Llama adapts via few-shot patterns, and Mistral delivers with minimal but precise phrasing. Use the table as a cross‑reference to rewrite prompts across ecosystems.

🔄 Rewrite the same prompt for each provider

Provider	Prompt structure (example skeleton)
OpenAI	Task: explain quantum computing → Context: high school students → Format: 3 paragraphs, bullet summary → Tone: enthusiastic & simple
Anthropic	[Role: expert tutor] [Background: student with basic physics] [Task: explain superposition] [Rules: avoid equations] [Output: analogies + real-world example]
Google/Gemini	Instruction: write a short educational note. Context: quantum bits and entanglement. Constraints: max 200 words. Output format: JSON with “explanation” and “key_terms”.
Meta (Llama)	Instruction: create an engaging explainer. Context: intro quantum mechanics → Constraints: use metaphors only → Examples: “like a spinning coin” → generate friendly style.
Mistral AI	Goal: teach superposition simply. Context: curious teenagers. Ask: write three short sentences and a one-line takeaway.

✨ The same core request reshaped for each model's “best prompt shape” shown in the mega table above.

⚙️ Unified reference — built from official prompting guides, community best practices, and benchmark insights. Updated for OpenAI, Anthropic Claude, Google Gemini, Meta Llama 3+, and Mistral Large/Mixtral.