AI Model Cost Comparator - Compare GPT, Claude, Gemini and Llama Prices

Paste a prompt and compare per-call, per-day and per-month cost across 12 AI models

🔒 100% di browser🆓 Gratis selamanya⚡ Tanpa signup
Catatan: Interface tool di bawah saat ini baru tersedia dalam bahasa Inggris. Kami sedang menerjemahkan UI satu per satu. Lihat tool ini dalam bahasa Inggris →
Estimated ~20 input tokens (rough: 4 chars/token).
Cheapest
GPT-4o mini · $0.000303/call
Most expensive
Claude Opus 4 · $0.0378/call
Spread
124.8x
Model$/1M in$/1M outPer callPer dayPer monthFits?
GPT-4o mini
OpenAI · 128k ctx
$0.15$0.6$0.000303$0.3030$9.09✓
Llama 3.3 70B
Meta (via Groq) · 128k ctx
$0.59$0.79$0.000407$0.4068$12.204✓
DeepSeek V3
DeepSeek · 64k ctx
$0.27$1.1$0.000555$0.5554$16.662✓
Gemini 2.5 Flash
Google · 1000k ctx
$0.3$2.5$0.001256$1.256$37.68✓
Claude Haiku 4.5
Anthropic · 200k ctx
$1$5$0.002520$2.52$75.60✓
Mistral Large 2
Mistral · 128k ctx
$2$6$0.003040$3.04$91.20✓
Gemini 2.5 Pro
Google · 1000k ctx
$1.25$10$0.005025$5.025$150.75✓
GPT-4o
OpenAI · 128k ctx
$2.5$10$0.005050$5.05$151.50✓
o1-mini
OpenAI · 128k ctx
$3$12$0.006060$6.06$181.80✓
Claude Sonnet 4.5
Anthropic · 200k ctx
$3$15$0.007560$7.56$226.80✓
o1 (reasoning)
OpenAI · 200k ctx
$15$60$0.0303$30.30$909.00✓
Claude Opus 4
Anthropic · 200k ctx
$15$75$0.0378$37.80$1,134.00✓
Prices are public list pricing per 1M tokens and change often. Token counts are estimated (~4 chars/token in English) - use exact input tokens for precise cost. Cached rate applies to Anthropic/OpenAI/DeepSeek prompt caching where supported.
Hanya estimasi. Hasil dihitung dari rumus umum berdasarkan input yang kamu masukkan, dan hanya untuk tujuan informasi. Bukan nasihat finansial, pajak, hukum, medis, atau profesional, dan tidak memperhitungkan situasi pribadi kamu, yurisdiksi, pajak, biaya, atau kondisi kesehatan. Selalu verifikasi angka penting dengan profesional yang berkualifikasi. Dengan memakai tool ini kamu setuju dengan Syarat & Disclaimer.

Cara pakai

  1. 1
    Paste your prompt

    The tool estimates input tokens (~4 characters per token in English). Enter exact token counts on the right for precise numbers.

  2. 2
    Set expected output tokens and call volume

    Roughly how long the responses will be, and how many calls you make per day.

  3. 3
    Adjust the cache hit rate

    If you use Anthropic, OpenAI or DeepSeek prompt caching, drag the slider up to see the discounted cost.

  4. 4
    Read the sorted table

    Models are sorted cheapest first, with per-call, per-day and per-month totals and whether the context window fits your prompt.

Tentang AI Model Cost Comparator - Compare GPT, Claude, Gemini and Llama Prices

Free AI model cost comparator. Paste your prompt or enter exact input tokens, set expected output length and calls per day, and see per-call and monthly cost across Claude Opus, Sonnet and Haiku, GPT-4o, o1, Gemini 2.5, Llama 3.3, Mistral Large and DeepSeek V3. Factor in prompt caching. Helps you pick the right model for your budget without spinning up 12 accounts. AI Model Cost Comparator - Compare GPT, Claude, Gemini and Llama Prices di 712 Tools jalan sepenuhnya di browser kamu pakai JavaScript API native - tidak ada server yang pernah melihat data kamu. Artinya hasil instan, privasi penuh, dan tanpa batas upload.

Buat sesi debug, quick sanity check, atau incident produksi - tool ini gratis dipakai sebanyak yang kamu butuh. Tanpa watermark, tanpa signup, dan tanpa ads di dalam tool.

Pertanyaan yang sering ditanyakan

How accurate are the token estimates?

The character-based estimate is within 10 to 15% for English text. Different models tokenize slightly differently, so for a precise budget, use the exact input token field or run a small sample through the real API.

Are these prices current?

Prices are public list prices at the time of the last update. Providers change pricing periodically; check the provider's pricing page before committing large volumes.

What is prompt caching?

Anthropic, OpenAI and DeepSeek offer a discount (roughly 90%) on input tokens that are repeated across calls, such as a long system prompt. If most of your prompt is a stable prefix, the cache hit rate is high and cost drops sharply.

Is my prompt uploaded anywhere?

No. All calculation runs in your browser. Your prompt text is not sent to any server.

Baca lebih lanjut