Every time you run the deep research feature on Claude, there's a real computational cost behind it. Tokens get processed, the model "thinks" through extended reasoning, web searches run, and a long answer gets assembled. That cost is invisible to the individual user because it's wrapped into one flat price, which is a monthly subscription. But once we try to translate that into an actual dollar figure, the picture gets interesting and that’s what this article digs into.
Foto oleh Google DeepMind di Unsplash
This piece tries to answer two questions: what does one deep research session actually cost when converted to API pricing, and what happens to that value when the resulting research is shared publicly instead of kept to yourself.
What Claude.ai Subscriptions Cost Right Now
Before getting into the math, it helps to align on the baseline pricing. Claude Pro is $20/month, and includes access to the latest models, extended reasoning, and MCP connectors. For users who need more headroom, there are two Max tiers: Max 5× at $100/month and Max 20× at $200/month, offering five and twenty times Pro's usage allowance respectively.
Outside of subscriptions, there's also the Claude API, billed per token. That's the yardstick we'll use to translate one research session into a concrete dollar figure.
Pricing Out a Single Deep Research Session
A deep research session isn't a single back-and-forth exchange. Several components contribute to the token count:
Component
Estimated Tokens
Notes
Initial input (prompt + context)
~10,000
Question and instructions
Web search & fetch
~20,000
Articles retrieved and processed by the model
Extended thinking
~30,000
Internal reasoning before answering
Final output
~5,000
Roughly 3,500–4,000 words of research output
Total
~65,000 tokens
For an Opus-class model, API pricing runs $15 per million input tokens and $75 per million output tokens. Based on the assumptions above, the rough math is:
Input (including web search and thinking): 60,000 tokens × $15/million ≈ $0.90
Output: 5,000 tokens × $75/million ≈ $0.375
Total per session: roughly $1.27
This is obviously an estimate and it could be lower with a Sonnet-class model, or higher for a much longer and more complex research task. But as a rough benchmark, one reasonably thorough deep research session is worth a little over a dollar if paid for directly through the API.
How Many Sessions Does One Subscription Actually Buy?
The next question: out of the $20/month paid for Pro, how many deep research sessions does that actually translate to?
This is a bit messy because Claude doesn't meter usage per message, it meters by computational load. A short conversation can consume a much smaller share of quota than a single deep research session involving many tool calls and long reasoning chains. The difference can run 10–20x. With a conservative assumption, an active Pro user running this feature regularly might get somewhere around 15–20 deep research sessions per month before hitting the limit.
Mapped to API-equivalent value:
Plan
Price/month
Est. Sessions/month
Equivalent API value
Difference
Pro
$20
~15–20
~$19–25
roughly break-even
Max 5×
$100
~75–100
~$95–127
modest savings
Max 20×
$200
~300–400
~$380–500
meaningful savings
The takeaway so far: from a pure compute-cost standpoint, the Pro subscription is roughly break-even against API pricing. The real value of subscribing isn't dollar savings, it's convenience: no API billing to manage, no token limits to track, no infrastructure to maintain.
The Turning Point: What Happens When Research Gets Shared
This is where the math gets more interesting. Up to this point, a single deep research session worth roughly $1.27 in compute cost. Typically stops with one person: whoever ran it. The output sits in a chat history, maybe gets copied into personal notes, and that's it.
Now picture a different scenario: that research gets published as a public article. Someone else who needs similar information no longer has to run the research from scratch, they just read it.
Run the numbers with conservative assumptions:
One Pro user runs 15 research sessions per month ($20)
Every result gets published as an article
Each article is read by an average of 500 people (a reasonable figure for a niche article shared within a community or on social media)
Without that article, some share of those 500 readers would likely have sought the same information another way — including running their own AI research
If all 500 of those readers had to run equivalent research independently through the API, the total cost would be 500 × $1.27 ≈ $635 per article.
Multiplied across 15 articles a month:
Metric
Value
Subscription cost
$20/month
Articles published
15/month
Readers per article (estimate)
500 people
Compute value per research session
$1.27
Total potential value saved for the community
~$9,525/month
Ratio vs. subscription cost
~476×
That 476× figure is obviously not a precise calculation, it depends entirely on assumptions about reader count and how relevant the research is to them. But the underlying direction holds: the value of one research session doesn't disappear after the first person reads it. Once published, the same value can be reused by different people, with no additional compute cost for each subsequent reader.
Why This Matters
There's a real access gap in how AI gets used today. Not everyone can afford a $20/month subscription, let alone $100–200 for a Max tier. Not everyone has the time or the skill to write a good research prompt either. But once one person has done that research well and shared it openly, others without the same access can still benefit from it for free, without needing to understand how the AI works at all.
This isn't a new argument. Libraries, open-access journals, and Wikipedia all run on the same logic. But with AI capable of producing deep research in minutes, the scale and speed at which knowledge could be distributed this way is potentially much larger than before. Provided there's a channel that makes "research on AI, then publish publicly" as frictionless as possible.
How to Actually Share It: The Frasa Connector
All of this math is interesting in theory, but it only matters if sharing research is actually easy. If publishing a deep research session requires copy-pasting into a separate publishing tool, reformatting, and manually posting it somewhere, most people simply won't bother and the 476x multiplier stays theoretical.
This is the gap the Frasa connector for Claude.ai is built to close. Frasa is a custom MCP (Model Context Protocol) connector that lets Claude publish directly to Frasa.io. Turning a finished research session into a public article without leaving the chat.
Setup takes three steps:
Open Custom Connectors in claude.ai settings
Enter the Frasa MCP server (api.frasa.io/mcp)
Connect — done. Claude can now publish to Frasa on request
Once connected, the workflow looks like this:
Run deep research in Claude → ask Claude to publish it via Frasa → it's live as a public article on Frasa.io
With the connector, there are no extra steps. You don't copy, reformat, or repost anything. The gap between "I just learned something useful" and "other people can now read it" basically disappears.
The connector doesn't change the math from before. It just makes that math real. The 476x multiplier only happens if sharing is as easy as the research itself, and this is what makes it that easy.
So the earlier numbers come down to one simple point. The question was never whether research is worth sharing, because it clearly is. The hard part was always the effort it took to share it. When you drop that effort close to zero, sharing stops being a rare one-time thing and becomes a normal habit for thousands of people.
wibowo•Jun 20, 2026Disclaimer: Harga produk AI generatif sangat dinamis dan dapat berubah sewaktu-waktu. Seluruh angka pricing dalam laporan ini dikumpulkan dari halaman resmi vendor dan sumber tracker pricing yang diakses pada akhir April – awal Mei 2026. Konversi USD ke IDR menggunakan kurs referensi USD 1 ≈ IDR 17.330 (rata-rata mid-market awal Mei 2026 — sumber: Xe, Trading Economics). Kurs ini sedang dalam tren pelemahan rupiah (USD/IDR sempat menyentuh 17.387 pada akhir April 2026). Indonesia adalah pasar AI generatif terbesar ke-5 dunia menurut traffic ChatGPT (≈3,95% traffic global), dan salah satu pasar dengan pertumbuhan tercepat (≈85% YoY menurut data ChatGPT Country Statistics Februari 2026). Dari 229,4 juta pengguna internet Indonesia (APJII, 2025), survei PwC Hopes & Fears 2025 menunjukkan 69% responden Indonesia pernah menggunakan AI generatif dalam 12 bulan terakhir dan 16% memakainya setiap hari — di atas rata-rata global (54% dan 14%). Namun, kesenjangan akses ke frontier models (model terdepan) sangat lebar. Subscription tier termurah yang sudah memberi akses Deep Research dan model terbaru (Claude Opus 4.7, GPT-5.5, Gemini 3.1 Pro, Perplexity Sonar Deep Research) berkisar USD 19,99–20 per bulan — setara Rp 346.000–347.000 , atau sekitar 6% dari UMP Jakarta 2026 (Rp 5.729.876) , dan >10% dari upah rata-rata pekerja Indonesia (Rp ~3 juta/bulan menurut Antom/BPS). Tier yang membuka kapasitas riset profesional sesungguhnya — ChatGPT Pro USD 200, Claude Max USD 200, Google AI Ultra USD 249,99, Perplexity Max USD 200 — berharga Rp 3,46 juta – 4,33 juta/bulan , atau 60%–76% dari UMP Jakarta . Bagi mayoritas mahasiswa, pekerja UMKM, dan jurnalis Indonesia, angka ini secara faktual prohibitive . Hipotesis utama riset ini didukung oleh data : terdapat cost gap nyata antara harga frontier AI dan daya beli rata-rata Indonesia, sehingga berbagi hasil deep research model-model ini ke publik (seperti yang dilakukan Frasa.io ) memiliki nilai redistribusi pengetahuan yang signifikan . Vendor Model Frontier Aktif Tanggal Rilis Catatan Anthropic Claude Opus 4.7 16 April 2026 Tokenizer baru menghasilkan hingga 35% token lebih banyak dari Opus 4.6; harga per-token tidak berubah. Sonnet 4.6 dan Haiku 4.5 tetap aktif. OpenAI GPT-5.5 dan GPT-5.5 Pro Tier reguler & Pro GPT-5.4 dan varian (5.4 mini/nano) adalah tier menengah; GPT-5.3 tersedia di tier gratis. Google Gemini 3.1 Pro (flagship), Gemini 3 Pro Deep Think Februari 2026 “Gemini Advanced” telah di-rebrand menjadi Google AI Pro sejak Maret 2026. Perplexity Sonar Pro / Sonar Reasoning Pro / Sonar Deep Research — Tidak memiliki model frontier sendiri; mengorkestrasi GPT-5.5, Claude Opus 4.7, Gemini 3.1 Pro melalui Agentic Research API. 229.428.417 pengguna internet di Indonesia, dari populasi 284,4 juta jiwa. Penetrasi internet 80,66% (naik 1,16 pp dari 79,5% di 2024). Penetrasi tertinggi di Pulau Jawa 84,69% ; terendah Maluku & Papua 69,26% . Penetrasi urban 83,56% vs rural 76,96% — gap menyempit signifikan dibanding 2024. Demografi pengguna internet didominasi Gen Z (25,54%) , Milenial (25,17%) , dan Gen Alpha (23,19%) . Indonesia menyumbang ±3,95% traffic ChatGPT global → diestimasi ≈30–32 juta pengguna ChatGPT mingguan aktif (dari 800 juta WAU global per Februari 2026). Indonesia & Filipina berkontribusi 6,7% dari total traffic global ChatGPT — pasar terbesar ke-5 dunia. Growth rate Indonesia: ≈85% YoY — tertinggi di Asia Tenggara, didorong populasi muda dan penetrasi smartphone tinggi. World Bank: Indonesia, India, Thailand, China termasuk negara dengan adoption rate self-reported AI generatif tertinggi (>90%) di kalangan knowledge worker. PwC (812 responden Indonesia): 69% pernah pakai AI generatif (vs 54% global); 16% daily users (vs 14% global). Indikator Nilai (Rp) Nilai (USD @17.330) Sumber UMP DKI Jakarta 2026 5.729.876/bulan ≈USD 331 Kepgub DKI No. 1142/2025 Upah rata-rata pekerja Indonesia ±3.000.000/bulan ≈USD 173 Antom/BPS 2024 Pengeluaran rumah tangga rata-rata per kapita 1.500.000/bulan ≈USD 87 BPS via Antom Median gaji software engineer Indonesia (NodeFlair 2025) 9.000.000/bulan ≈USD 519 NodeFlair 2025 Range gaji software developer Jakarta 6,5 – 13,2 juta/bulan USD 375 – 762 Glassdoor Jan 2026 Kebutuhan Hidup Layak (KHL) keluarga kecil Jakarta (estimasi BPS via KSPI) ±15.000.000/bulan ≈USD 866 KSPI/BPS 2025 Penetrasi kartu kredit rendah. Visa/Mastercard mendominasi pembayaran lintas negara namun penetrasi kartu kredit Indonesia rendah — kartu kredit & debit hanya 17% volume e-commerce (GlobalData, Antom 2025). Dompet digital (GoPay, OVO, DANA, ShopeePay) dan QRIS adalah arus utama, tetapi sebagian besar layanan AI luar negeri tidak menerima e-wallet lokal , hanya kartu kredit/debit internasional, PayPal, atau Apple/Google Pay. Ini secara struktural mengeluarkan jutaan calon pelanggan. Harga ditagih dalam USD , sehingga depresiasi rupiah (USD/IDR melemah ±5,7% dalam 12 bulan terakhir per Mei 2026) terus menaikkan harga efektif tanpa kenaikan harga rate card. Bahasa : kualitas output Bahasa Indonesia masih lebih rendah dibanding Inggris (DesignRush 2026 mencatat performa ChatGPT optimal di Inggris/Spanyol/Prancis/Mandarin). Gap literasi digital & talenta : prediksi defisit 6–9 juta profesional IT sampai 2035 (AI Asia Pacific Institute, 2025). Awareness fitur premium : banyak pengguna Indonesia hanya mengenal free tier ; fitur Deep Research, Agent Mode, dan Codex sering tidak diketahui. Korporasi & UMKM : PwC Indonesia 2026 — daily GenAI users di Indonesia melaporkan 96% peningkatan produktivitas (vs 92% global), 82% rasa aman pekerjaan, 72% kenaikan gaji. Pendidikan : ±50% mahasiswa Indonesia menggunakan AI tools “beberapa kali sebulan” untuk tugas akademik (Statista AI in Indonesia 2024). Use case ChatGPT global yang paling populer (First Page Sage Apr 2026 + OpenAI): 80% interaksi adalah Practical Guidance, Seeking Information, dan Writing — dominan riset umum, riset akademik, content creation, coding, customer service. Studi kualitatif Frontiers/Telkom University (2025) mengidentifikasi tiga klaster pengguna ChatGPT Indonesia: early adopters , pengguna prompt engineering , dan pengguna kasual; AI dipersepsikan tidak hanya sebagai alat tetapi sebagai “mitra komunikatif”. Vendor Free Entry/Go Pro/Standard Max/Pro Tinggi Ultra/Enterprise Anthropic Claude Free (Sonnet/Haiku, ±10–15 msg/sesi) — Pro USD 20/bln (Opus 4.7 terbatas, Research, ±5× Free) Max USD 100 (5×) atau USD 200 (20×) Team USD 25–125/seat ; Enterprise mulai USD 20/seat + API terpisah OpenAI ChatGPT Free (GPT-5.3, 10 msg/5 jam, berisi iklan di US ) Go USD 8/bln (10× free, no GPT-5.5 chat) Plus USD 20/bln (GPT-5.5 Thinking, 10 Deep Research/bulan) Pro USD 100 (5×) atau USD 200 (20×, 250 Deep Research/bulan, 1M context) Business USD 20–25/seat (min 2 user); Enterprise custom Google Gemini Free (Gemini 3 Flash, ±5 Deep Research/bulan) AI Plus USD 7,99/bln AI Pro USD 19,99/bln (Gemini 3.1 Pro, Deep Research, 1.000 AI credits, 2 TB) — AI Ultra USD 249,99/bln (Deep Think, Veo 3.1, Agent Mode, 25.000 credits, 30 TB, $100 GCP credit) Perplexity Free (terbatas, no Deep Research) — Pro USD 20/bln (unlimited Pro queries; ±20 Deep Research/hari) Max USD 200/bln (Comet Max, video gen) Enterprise Pro USD 40/seat ; Enterprise Max USD 325/seat Tier USD/bulan IDR/bulan % UMP Jakarta Setara harga lokal ChatGPT Go 8 138.640 2,4% ≈4 cup kopi specialty Jakarta (Rp 35–40 rb) Google AI Plus 7,99 138.470 2,4% ≈1 hari makan kelas menengah ChatGPT Plus / Claude Pro / Perplexity Pro / Google AI Pro 19,99–20 ≈346.600–347.000 6,1% ±1 hari kerja UMR (Rp 191 rb/hari) Claude Max 5× / ChatGPT Pro $100 100 1.733.000 30,2% ≈setara cicilan motor entry-level Claude Max 20× / ChatGPT Pro $200 / Perplexity Max 200 3.466.000 60,5% >60% UMP Jakarta Google AI Ultra 249,99 4.331.700 75,6% ±gaji UMR di provinsi NTT/Jateng Perplexity Enterprise Max (per seat) 325 5.632.250 98,3% Hampir setara 1× UMP Jakarta Catatan kontekstual : Secangkir kopi spesialti di kafe Jakarta = Rp 35.000–55.000. Sehari makan layak (3× makan sederhana) = Rp 60.000–100.000. UMP Jakarta dibagi 22 hari kerja = Rp ±260.000/hari kerja. Dengan demikian, Claude Pro ≈ 1 hari kerja UMR penuh , dan Google AI Ultra ≈ 16 hari kerja UMR . Vendor Tier yang Memberi Deep Research Limit Claude (Anthropic) Pro, Max, Team, Enterprise Tidak dipublikasi; 5-hour rolling window, 7-day weekly cap; pengguna Pro biasanya kena limit dalam 1–2 jam pemakaian intensif (CheckThat 2026) ChatGPT Plus, Pro, Business, Enterprise Plus = 10 Deep Research/bulan ; Pro $200 = 250/bulan ; Free = “limited” Gemini Free, AI Plus, AI Pro, AI Ultra Free = ±5/bulan ; Pro/Ultra = “Higher/Highest” tanpa angka pasti Perplexity Pro, Max, Enterprise Pro = ±20 Deep Research/hari (ScreenApp Feb 2026) — paling generous untuk individu Model Input (USD/MTok) Output (USD/MTok) Konteks Maks Claude Opus 4.7 5,00 25,00 1 M token (flat rate) Claude Sonnet 4.6 3,00 15,00 1 M token Claude Haiku 4.5 1,00 5,00 200 K GPT-5.5 ±5,00 ±15,00 (kompilasi CloudZero/Finout) — GPT-5.4 2,50 15,00 — Gemini 3.1 Pro 2,00 12,00 ≤200 K (di atas 200 K: 2× rate) Gemini 2.5 Pro 1,25 10,00 ≤200 K Gemini 3 Flash-Lite 0,25 1,50 — Perplexity Sonar Deep Research 2,00 (input) + 2,00 (citation) + 3,00 (reasoning) 8,00 + USD 5/1.000 search query Perplexity Sonar Pro 3,00 15,00 + USD 6–14/1.000 request Perplexity Sonar (base) 1,00 1,00 + USD 5–12/1.000 request Input tokens : 150.000 – 400.000 (akumulasi konten yang di- scrape + system prompt + context conversation) Output tokens (termasuk thinking) : 30.000 – 80.000 Web searches : 20–40 Skenario API Estimasi Biaya/Sesi (USD) IDR (Rp 17.330) Claude Opus 4.7 (250 K input / 50 K output, no caching) (0,25 × 5) + (0,05 × 25) = 2,50 ≈43.300 Claude Opus 4.7 (dengan 70% cache hit) ≈ 1,15–1,40 ≈20.000–24.300 GPT-5.5 (250 K / 50 K) (0,25 × 5) + (0,05 × 15) = 2,00 ≈34.660 Gemini 3.1 Pro (200 K / 50 K) (0,2 × 2) + (0,05 × 12) = 1,00 ≈17.330 Perplexity Sonar Deep Research (250 K / 50 K + 30 search) (0,25 × 7) + (0,05 × 8) + (30/1000 × 5) = 2,30 ≈39.860 ChatGPT Plus USD 20 ÷ 10 Deep Research = USD 2,00/sesi (Rp 34.660) ChatGPT Pro USD 200 ÷ 250 Deep Research = USD 0,80/sesi (Rp 13.860) — paling efisien per-sesi untuk power user Perplexity Pro USD 20 ÷ ±600 Deep Research/bulan (20/hari) = USD 0,033/sesi (Rp 575) — sangat agresif, namun kualitas hasil bervariasi Claude Pro USD 20: tidak ada angka resmi; pengguna power biasanya hanya dapat 5–15 sesi panjang/bulan = USD 1,30–4,00/sesi Satu deep research session yang mengkonsumsi ±300 K token (input+output) pada model kelas frontier (≥400 B params, MoE-hybrid) diperkirakan setara 0,4–1,5 kWh energi, atau 0,2–0,8 kg CO₂-eq tergantung mix energi data center (sumber pembanding: Hugging Face AI Energy Score 2025; angka indikatif). Sebagai perbandingan: 1 kWh ≈ 1 jam menyalakan AC 1 PK; 0,5 kg CO₂ ≈ ±2 km berkendara mobil bensin. Klaim “hours of GPU work” yang sering dipromosikan vendor (mis. ChatGPT Pro: “satu Deep Research = 30 menit kerja PhD”) belum diverifikasi independen dan harus dibaca sebagai marketing language . Cost gap nyata. Tier “deep research siap pakai” (USD 20/bulan ≈ Rp 347 ribu) memang tidak prohibitive untuk kelas menengah-atas urban, tetapi: Mewakili ≈12% dari upah rata-rata Indonesia (Rp 3 juta) — di atas threshold “discretionary spending” 5% yang dipakai Bank Dunia. Mewakili >50% dari UMP provinsi terendah (Jawa Tengah Rp 2,3 juta). Tier Pro/Ultra (USD 100–250) secara faktual hanya terjangkau senior knowledge worker urban dengan gaji ≥Rp 15 juta/bulan — segmen yang menurut LinkedIn Talent Insights kurang dari 5% angkatan kerja Indonesia. Hambatan non-harga lebih besar dari sekadar harga : kartu kredit penetration rendah, pricing dalam USD yang volatile, kualitas Bahasa Indonesia di model masih inferior, dan literasi prompt engineering yang tidak merata. Konteks regional memperjelas posisi Indonesia: Negara ASEAN GDP per kapita 2025 (USD, IMF) Harga Plus/Pro USD 20/bulan = % GDP/kapita bulanan Rasio relatif Singapore 89.000 0,27% Sangat terjangkau Malaysia 13.500 1,78% Terjangkau Thailand 7.800 3,08% Cukup terjangkau Indonesia 5.500 4,36% Marginal Vietnam 4.700 5,11% Marginal Filipina 4.200 5,71% Marginal Pada Pro/Ultra USD 200, persentase Indonesia melonjak ke 43,6% dari pendapatan bulanan rata-rata — tidak masuk akal untuk individu kebanyakan. Stratifikasi pengetahuan baru : pekerja yang mengakses Opus 4.7/GPT-5.5 Pro berpotensi menghasilkan riset, kode, dan konten 20–70% lebih cepat (PwC 2026, McKinsey 2025) — memperkuat winner-takes-most dynamics di pasar tenaga kerja Indonesia. Kesenjangan UMKM vs korporasi : korporasi besar membayar Enterprise tier; UMKM (61% PDB Indonesia) tertinggal di free tier dengan model lebih lemah dan tanpa Deep Research. Ketergantungan platform asing : 100% frontier models yang dianalisis di-host di luar Indonesia — implikasi pada data sovereignty , latency , dan biaya transfer pajak digital. Peluang positif : NBER (2025) menemukan adoption rate AI di negara berpendapatan terendah tumbuh >4× lebih cepat dibanding negara kaya — Indonesia ada di sweet spot pertumbuhan ini, sehingga intervensi redistribusi pengetahuan punya leverage tinggi. Publikasikan hasil deep research dengan metadata transparan — sebutkan model yang dipakai (mis. “Hasil Deep Research dengan Claude Opus 4.7 + Perplexity Sonar Deep Research, 4 Mei 2026”), prompt utama, dan jumlah sumber yang dirangkum. Ini meningkatkan epistemic trust . Sertakan terjemahan & lokalisasi . Output mentah AI cenderung Inggris-sentris; melakukan post-editing ke Bahasa Indonesia dengan konteks lokal (regulasi, statistik BPS, contoh UMKM) meningkatkan utilitas masyarakat luas. Open-source prompt & metodologi . Bagikan research recipe , bukan hanya hasil — ini memperkuat literasi prompt engineering komunitas. Bangun kurasi tematik sektor strategis : pertanian, perikanan, kesehatan ibu-anak, UMKM, pendidikan vokasi — sektor yang ROI sosialnya tinggi tapi underserved oleh konten AI berbahasa Inggris. Gunakan model Free + frontier hybrid : free tier (Gemini Free, ChatGPT Free, Claude Free, Perplexity Free) dapat menangani 70–80% kebutuhan; gunakan Deep Research berbayar hanya untuk high-leverage queries yang hasilnya akan dipublikasikan. Advokasi pricing regional / student tier . Google sudah memberi 1 tahun gratis Google AI Pro untuk mahasiswa terverifikasi; Anthropic memiliki university plan; OpenAI memberikan ChatGPT Edu untuk universitas. Komunitas Indonesia perlu mendorong tier pricing PPP-adjusted seperti yang sudah dilakukan Spotify, Netflix, dan YouTube Premium. Platform demokratisasi seperti Frasa.io — yang mengkurasi & menerjemahkan hasil deep research frontier ke publik Indonesia — mengisi market failure nyata: pengguna yang tidak mampu atau tidak punya akses pembayaran USD tetap memperoleh manfaat frontier-grade insights . Pricing snapshot : data dari halaman resmi vendor ( claude.com/pricing , chatgpt.com/pricing , gemini.google/subscriptions, docs.perplexity.ai/docs/getting-started/pricing ) dan tracker pihak ketiga (Finout, IntuitionLabs, MetaCTO, AI Pricing Guru, CheckThat, CloudZero, CometAPI), diakses akhir April – awal Mei 2026. Data demografi : APJII Profil Internet Indonesia 2025 (rilis 6 Agustus 2025), Statista AI in Indonesia, PwC Hopes & Fears 2025 (rilis 23 Februari 2026), World Bank/NBER “Who on earth is using Generative AI” (2024–2025), Siana ChatGPT Country Statistics Februari 2026, First Page Sage April 2026. Ekonomi : BPS, Trading Economics (USD/IDR mid-market 4 Mei 2026 = Rp 17.368,7), Kepgub DKI No. 1142/2025 (UMP Jakarta 2026), NodeFlair Salary Report 2025, Glassdoor Jakarta Januari 2026. Caveat utama : Beberapa angka adopsi AI di Indonesia (mis. “16% daily users”) berasal dari survei self-report dengan sampel terbatas (PwC: 812 responden Indonesia) — perlakukan sebagai indikator arah, bukan presisi statistik. Estimasi token & biaya per deep research session adalah rentang konservatif berbasis pengalaman komunitas pengembang; biaya aktual dapat lebih tinggi pada query agentic kompleks (multi-step, multi-modal). Carbon footprint model frontier belum dilaporkan resmi oleh Anthropic/OpenAI/Google/Perplexity untuk inferensi per-request; angka di Bagian 4.5 adalah ekstrapolasi indikatif dari literatur akademik. Kurs rupiah dalam tren melemah; harga IDR akan naik di tier yang sama jika USD/IDR menembus 17.500–18.000 pada paruh kedua 2026 (proyeksi Long Forecast & Bank Indonesia).
FrasaToday•May 4, 2026