What's new in July 2026
July was about making Twelve Data as easy for machines to read as it is for people. Alongside that: a page for every exchange in the world, wider mic_code support, and a solid round of fixes.
Built for AI agents
An agent can now find its way around Twelve Data without a human pointing the way.
We publish llms.txt and llms-full.txt as plain text files, so any LLM or agent can ingest our API surface directly instead of parsing HTML pages built for humans. The site also serves a machine readable card for our MCP server, a catalog entry pointing straight at the OpenAPI spec, and an authentication guide written for agents rather than people. Point a tool at twelvedata.com and it can work out what we offer, how to sign in, and where to send requests on its own.
The documentation has a new home for all of it. What used to be Artificial Intelligence is now AI & Agents, covering the MCP Server, integrations for ChatGPT, OpenClaw and NEAR Agent, llms.txt, and the World Model Dataset.
A page for every exchange
There is now a directory covering every exchange we know of, including ones we do not carry data for yet. Each exchange gets its own page with the full name, MIC and operating MIC, country, market name, institution description, and market category code.
Useful if you are mapping instruments across venues, checking which MIC belongs to which operator, or just want to see how wide the world of listed markets actually is. Start at twelvedata.com/exchanges.
API updates
mic_codeacross more endpoints. Exchange disambiguation now works consistently outside price data, so endpoints like price target accept the same MIC based filtering you already use elsewhere.- Crypto defaults to Binance. When you request crypto pricing without naming an exchange, Binance is now the top ranked source.
/quotereturns correct previous close. Some instruments were returning a wrongprevious_close, which threw offchangeandpercent_changeas well. Both are accurate now.- Symbol search finds more Australian instruments. A handful of active symbols on Australian venues only turned up in searches that included deleted listings. They now appear in normal results.
- Clearer errors. Requests for identifiers with no exchange attached return a readable message instead of a vague failure.
Richer reference data
Company logos are now collected automatically from issuer websites, so logo coverage keeps up as new companies list. We have also synced CIK codes for US tickers, which makes cross referencing against SEC filings much less painful.
Account and billing
- Switch between personal and business plans yourself, in both directions, as long as your payment method stays the same
- Reset your password from the sign in page without emailing support
- Reviewed and cleaned up the integrations settings page
See what is included in each plan at twelvedata.com/pricing.