Question 1

What is the cheapest AI API in March 2026?

Accepted Answer

Gemini 3.1 Flash is the cheapest at $0.15/M input tokens and $0.60/M output tokens. DeepSeek V4 is the next cheapest at $0.55/$2.19. For cached tokens, Gemini Flash drops to just $0.04/M.

Question 2

How much does Claude Opus 4.6 cost?

Accepted Answer

Claude Opus 4.6 costs $15/M input tokens and $75/M output tokens. With prompt caching, input costs drop to $7.50/M. It offers the best reasoning quality but at premium pricing.

Question 3

What is cached token pricing?

Accepted Answer

Cached token pricing applies when you send the same prompt prefix repeatedly. Providers store this prefix and charge 50-75% less for repeated context. This is especially valuable for system prompts and few-shot examples.

Question 4

GPT-5.4 vs Claude Sonnet 4.6 - which is better value?

Accepted Answer

Claude Sonnet 4.6 ($3/$15 per M tokens) is cheaper than GPT-5.4 ($5/$15) for input-heavy workloads. GPT-5.4 matches on output pricing. Sonnet also offers better cached pricing at $1.50/M vs $2.50/M.

Model	Input	Input Cache	Output	Konteks	Kecepatan
GPT-5.4 OpenAI	$5.00	$2.50	$15.00	256K	Medium
Claude Opus 4.6 Anthropic	$15.00	$7.50	$75.00	1M	Slower
Claude Sonnet 4.6 Anthropic	$3.00	$1.50	$15.00	200K	Fast
Gemini 3.1 Pro Google	$2.50	$0.63	$10.00	2M	Medium
Gemini 3.1 Flash Google	$0.15	$0.04	$0.60	1M	Fastest
DeepSeek V4 DeepSeek	$0.55	$0.14	$2.19	128K	Fast

Cek Harga API

Volume Bulanan

Perbandingan Biaya Bulanan

Kartu Tarif Per-Token (per 1 juta token)

Integrate Any AI API with ZBuild

Related Tools