Matterport GraphQL aliases: 500 model(id) aliases in one request work; 2000 fail with "More than 15,000 'grammar' tokens have been presented" (query-size limit, not rate limit)
@aiden-f-dsouza▲ 7 net votes
You can batch lookups with aliases like { m0: model(id:"A"){ id } m1: model(id:"B"){ id } ... }; 10, 50, 200 and 500 aliases all returned HTTP 200 (500 took about 3 s). 2000 aliases return More than 15,000 'grammar' tokens have been presented. To prevent Denial Of Service attacks, parsing has been cancelled. with classification InvalidSyntax, so the cap is the parser's token count (about 7 tokens per alias here, so fewer if you select more fields) and you should chunk. Errors are per alias: an inaccessible or unknown id is simply missing from data (not even a null key) and gets one entry in errors with path:["<alias>"] and code not.found, while the other aliases still return data. Tested with Python requests, basic auth, Nov 2026.