When is pricing updated?

Refreshed at the start of each quarter based on public vendor pricing.

How much does caching save?

Cached input tokens cost 25-50% of normal. 50%+ cache ratio saves 20-40%.

How is TR quality calculated?

Internal calibration (Q1 2026): MMLU-TR, TruthfulQA-TR, content quality, TR ecosystem benchmark.

What are reasoning tokens?

Extra tokens reasoning models (o-series, DeepSeek-R) produce during 'thinking'; priced separately.

A lot — KVKK Art. 9 cross-border compliance, AI Act jurisdiction and DPA availability depend on region.

Are open-source models really cheap?

Hosted (Together AI, Fireworks) is low; self-hosted should include GPU CapEx + electricity + engineer cost.

AI Interactive Tools

LLM Token Cost Comparator

18+ LLM models with Q1 2026 pricing + prompt caching + reasoning + TR quality + region filter.

Definition

LLM Token Cost: The fee LLM models charge per 1M input + 1M output tokens. Monthly total: ((input × $/M_in) + (output × $/M_out)) × monthly requests.; Also known as: input pricing, output pricing, prompt caching, $/1M token

Inputs

Monthly RequestsAvg Input TokenstokAvg Output TokenstokReasoning TokenstokPrompt caching usedCache ratio: 50%

Filter

TierProvider

Models (5)

Comparison

Sign-up Required

LLM Token Cost Comparator (Q1 2026) results are members-only

You can adjust the form inputs freely; the result table, charts and PDF report require a free account. Your current inputs are preserved when you sign up.

Re-download your reports and PDFs from your dashboard
Stay updated on new tools and KVKK + EU AI Act changes
Full access to the Resource Centre, Forum and Learning Portal

KVKK/GDPR compliant — only name and email. We won't send ads; you can delete your account anytime.

Frequently Asked Questions

Refreshed at the start of each quarter based on public vendor pricing.

References

OpenAI Platform Pricing, OpenAI
Anthropic Pricing, Anthropic
Google AI Studio Pricing, Google
AWS Bedrock Pricing, AWS
Artificial Analysis Benchmarks, Artificial Analysis