How do I count tokens for GPT-5 or Claude?

Paste your text into the TokenCost counter. It uses the same tokenizers (cl100k_base and o200k_base) as OpenAI and calculates costs across 35+ models including GPT-5, Claude Opus, Gemini, and more.

Is there a free API for LLM pricing data?

Yes. TokenCost provides a free JSON API at tokencost.app/api/models with no authentication required, CORS enabled, and support for filtering by provider or model ID.

Which LLM models does TokenCost support?

TokenCost supports 35+ models from OpenAI (GPT-5.4, GPT-5, o3, o4-mini), Anthropic (Claude Opus, Sonnet, Haiku), Google (Gemini 3), xAI (Grok 4), Meta (Llama 4), Mistral, and DeepSeek.

Compare Models

Side-by-side pricing and specs. Pick two models, set your token volume, choose your currency.

Model A

Model B

Tokens

Currency

For 10,000 input tokens, GPT-5.2 is 65% cheaper than Claude Opus 4.6

Spec	GPT-5.2	Claude Opus 4.6
Provider	OpenAI	Anthropic
Input / 1M tokens	$1.75	$5.00
Output / 1M tokens	$14.00	$25.00
Input cost (10,000 tokens)	$0.018	$0.050
Output cost (10,000 tokens)	$0.140	$0.250
Context window	400K	200K
Max output	16K	32K

Cost at Scale (Input Tokens)

1,000 tokens$< 0.01$< 0.01

10,000 tokens$0.018$0.050

100,000 tokens$0.175$0.500

1,000,000 tokens$1.75$5.00

10,000,000 tokens$17.50$50.00

GPT-5.2Claude Opus 4.6

Exchange rates are approximate. Pricing as of March 2026. The green highlight indicates the better value for each metric.