AI model cost comparison

Paste once. Compare costs across six models.

Output tokens (including reasoning)Number of calls

Shared input assumption—

Have a token count? Adjust tokens or DeepSeek peak rates

Use a shared input token countUse DeepSeek off-peak rates (peak by default)

Peak: weekdays 01–04 and 06–10 UTC, excluding Chinese public holidays. Other hours are off-peak.

Cost comparison

Official API prices checked 2026-10-23 · USD / million tokens

Model / price source	Input rate	Output rate	Per call	Total estimate× 1,000
V4.1 Flash · 1017 ↗DeepSeek · Peak · cache miss	$0.30	$1.20	—	—
GLM-5.3 ↗Z.ai · Standard API	$1.40	$4.40	—	—
Qwen3.8 Max ↗Alibaba Cloud · International · Singapore	$2.00	$6.00	—	—
GPT-6.1 Sol ↗OpenAI · Standard · base context tier	$2.00	$10.00	—	—
Claude Sonnet 5.5 ↗Anthropic · Standard · global endpoint	$2.00	$10.00	—	—
Gemini 3.8 Flash ↗Google · Promotion through 2026-12-31	$0.75	$3.75	—	—

Paste content to see all six cost estimates here.

Token counts differ across models. This shared rough estimate assumes 4 ASCII characters or 1 other character per token; you can enter a token count instead. This compares budgets, not equivalent answer quality.

Billing assumptions and privacy

Per-call cost = (input tokens × input rate + output tokens × output rate) / 1,000,000. Multiply by calls for the total. Output should include billed reasoning tokens.

Uses standard text API cache-miss rates. Free quotas, batch discounts, search tools, image/audio, taxes and regional surcharges are excluded. Gemini switches to its announced new rates on 2027-01-01. The calculator scope is not a model context-capacity guarantee.

Content is processed only in this browser: no upload, storage or model calls. Switching tools retains input; reloading clears it. CSV exports contain usage and costs, never the pasted text.