AI Model API Pricing Comparison (per Million Tokens)

Explore the AI API market: compare 4 models across 13 channels and 6 providers. Compare input, output and cached token prices per million tokens before estimating total cost.

Price data updated on: 2026-11-09 (UTC)

Read pricing guides View the API Price Index

models

API channels

providers

China-reachable

Filters

Limited-time offers only

0 models matched "qwen-qwen3-8-max-0902-batch"

Model Channel Prices

No results found. Try adjusting your filters.

Frequently asked questions

How do I compare AI model API token prices?

This table covers 4 models and 13 available API channels from 6 providers. Compare input, output and cached input prices separately per million tokens; your cost depends on the amount of each token type used.

Which APIs have the lowest input token prices?

Among available channel headline prices, the lowest rates per million input tokens in each currency are: gemini-3.8-flash (XycAi) at CNY 2.5875; Hy4 (OpenRouter) at $0.417. Check output, cached rates and other price variants separately.

Can I directly compare API prices in different currencies?

Compare amounts in the same currency first. Quotes retain the provider’s original currency, and the lowest input rates above are listed separately by currency. Cross-currency comparisons require an exchange rate check.

When was the API pricing data updated?

The latest database price-row update among available LLM channels and their public price variants is 2026-11-09 (UTC). This does not mean every channel was verified that day; confirm the provider’s quote before purchasing.

Web search & fetch tool pricing (per request)

Official per-request prices; token costs are separate. Each row links its official source and verification date.

Provider	Tool	Per-request price	Token	Notes	Source
Anthropic	Web search	$10 / 1,000 searches	Plus search content tokens at model rates	Each search counts once regardless of results; errors are not billed.	Official 2026-10-18
Anthropic	Web fetch	No per-call fee	Fetched content billed as input tokens	No additional charge beyond standard token costs.	Official 2026-10-18
OpenAI	Web search	$10 / 1,000 calls	Plus search content tokens at model rates	All models; gpt-4o-mini / gpt-4.1-mini bill a fixed 8,000 input-token block per call.	Official 2026-10-18
OpenAI	Web search preview (non-reasoning models)	$25 / 1,000 calls	Search content tokens are free	—	Official 2026-10-18
Google	Grounding with Google Search (Gemini 3+)	$14 / 1,000 search queries	Retrieved context not billed as input	5,000 free requests/month shared across Gemini 3+ models; one prompt may trigger several queries.	Official 2026-10-18
Google	Grounding with Google Search (Gemini 2.5)	$35 / 1,000 grounded prompts	Retrieved context not billed as input	1,500 requests/day free on the paid tier.	Official 2026-10-18
Google	URL context	No per-call fee	Fetched content billed as input tokens	—	Official 2026-10-18
xAI	Web search	$5 / 1,000 calls	Plus tokens at model rates	Includes browsing pages and image search.	Official 2026-10-18
DeepSeek	Search API	¥5 / 1,000 successful requests	Standalone API, no tokens	US$0.75 on the English page. Server-side calls from Claude Code / Codex are free until 2026-11-01 00:00 Beijing time.	Official 2026-10-18
DeepSeek	Fetch API	¥5 / 1,000 successful requests	Standalone API, no tokens	US$0.75 on the English page (https://api-docs.deepseek.com/quick_start/pricing). Same free period as Search.	Official 2026-10-18
Vercel AI Gateway	DeepSeek Search	$0.75 / 1,000 billable requests	Host model tokens billed separately	Gateway channel, not a new DeepSeek list price. Docs table: $0.75 per 1,000 billable requests, the same figure as the DeepSeek English page. Billable when a successful request returns a nonempty results array, including results filtered before they reach the model. Failed requests, timeouts, and empty results are not billed. Result count does not change the price. Changelog: https://vercel.com/changelog/ai-gateway-adds-deepseek-search-and-fetch-tools	Official 2026-11-07
Vercel AI Gateway	DeepSeek Fetch	$0.75 / 1,000 billable requests	Host model tokens billed separately	Gateway channel. Docs table: $0.75 per 1,000 billable requests. Billable when a successful request returns nonempty page content. Failed requests, timeouts, and empty results are not billed. Changelog: https://vercel.com/changelog/ai-gateway-adds-deepseek-search-and-fetch-tools	Official 2026-11-07
Baidu AI Cloud	Baidu Search API (Enhanced / 增强版)	¥40 / 1,000 calls	Standalone API, no tokens	Domestic content list price: ¥0.04/call. Limited-time campaign 2026-10-30 to 2026-12-31: ¥4.9 per 1,000 calls (prepaid pack, up to 1M calls cumulative; overage billed at list price). 1,500 free calls/month; default 3 QPS. Overseas content: ¥0.072/call. Campaign: https://cloud.baidu.com/product/AI-SEARCH/campaign.html	Official 2026-10-31
Baidu AI Cloud	Baidu Search API (Flagship / 旗舰版)	¥72 / 1,000 calls	Standalone API, no tokens	Domestic content list price: ¥0.072/call. Limited-time campaign 2026-10-30 to 2026-12-31: ¥9.9 per 1,000 calls (invite code required). 1,500 free calls/month; default 3 QPS. Overseas content: ¥0.072/call. Campaign: https://cloud.baidu.com/product/AI-SEARCH/campaign.html	Official 2026-10-31