AI API Price Comparison
Every price below was read from the provider's own pricing page — never from another calculator site — and each row links back so you can verify it yourself in one click.
Model price comparison
Standard API tier, USD per 1 million tokens. Prices checked 30 Oct 2026 — verify on the provider site before you budget, because prices change.
| Model | Provider | Input / 1M | Output / 1M | Official pricing |
|---|---|---|---|---|
| gpt-6-astra | OpenAI | $10.00 | $50.00 | Verify ↗ |
| gpt-6.1-sol | OpenAI | $2.00 | $10.00 | Verify ↗ |
| gpt-6.1-terra | OpenAI | $1.00 | $5.00 | Verify ↗ |
| gpt-6-luna | OpenAI | $0.10 | $0.50 | Verify ↗ |
| gpt-5.3-codex | OpenAI | $1.75 | $14.00 | Verify ↗ |
| Claude Opus 4.8 | Anthropic | $5.00 | $25.00 | Verify ↗ |
| Claude Sonnet 4.6 | Anthropic | $3.00 | $15.00 | Verify ↗ |
| Claude Haiku 4.5 | Anthropic | $1.00 | $5.00 | Verify ↗ |
| Claude Haiku 3.5 | Anthropic | $0.80 | $4.00 | Verify ↗ |
| Gemini 4 ArgonPromotional rate listed through 30 Nov 2026; page lists $4.00 / $20.00 from 1 Dec 2026. | $2.00 | $10.00 | Verify ↗ | |
| Gemini 3.1 Pro PreviewPrompts up to 200k tokens; higher rate applies above 200k. | $2.00 | $12.00 | Verify ↗ | |
| Gemini 3.8 FlashRate listed through 31 Dec 2026; page lists $1.50 / $7.50 from 1 Jan 2027. | $0.75 | $3.75 | Verify ↗ | |
| Gemini 2.5 ProPrompts up to 200k tokens; higher rate applies above 200k. | $1.25 | $10.00 | Verify ↗ | |
| Gemini 2.5 FlashText price; audio input is priced higher. | $0.30 | $2.50 | Verify ↗ | |
| Gemini 3.1 Flash-LiteText / image / video price; audio input is priced higher. | $0.25 | $1.50 | Verify ↗ |
Every row: prices checked 30 Oct 2026 — verify on provider site. Tier notes (long-context, promotional windows, audio rates) are shown under the model name where the official page lists them.
How to read this table
Prices are the standard, pay-as-you-go tier per 1 million tokens. Most providers also offer cheaper batch modes, cached-input rates and enterprise discounts that are not shown here, because they depend on how you call the API. Where a price only applies up to a context size, or is a promotional rate with a published end date, the note under the model name says so. When two models cost the same on paper — as several mid-tier models currently do — the deciding factors become output quality on your own test set, latency and context length, not price.