diff --git a/src/util/make_model.js b/src/util/make_model.js
index 111e9e4..5db027a 100644
--- a/src/util/make_model.js
+++ b/src/util/make_model.js
@@ -36,7 +36,8 @@ export default (options={}) => {
 	// private model cache & meta-info
 	let _cache = {
 		data: null,
-		count: null
+		count: null,
+		upstream_limit: null
 	}
 
 	// some random places to make things work
@@ -49,10 +50,20 @@ export default (options={}) => {
 		cache() { return _cache },
 
 		// main methods
-		select(from=0, to) {
+		select(offset=0, limit=Infinity) {
 			// initialize data (singular or plural)
 			_cache.data = _cache.data || []
 
+			// normalize limit
+			if (limit == Infinity)
+				limit = _cache.upstream_limit || _cache.count || Infinity
+
+			// be lazy: if the data is presented, return the value immediately
+			//
+			// note that Array.every(list) == true if list is empty
+			if (_cache.data.slice(offset, offset+limit).length > 0 && _cache.data.slice(offset, offset+limit).every(data => data != undefined))
+				return Promise.resolve(_cache.data.slice(offset, offset+limit))
+
 			// now the hard work
 			return _configs.api.request({
 				method: 'GET',
@@ -65,20 +76,29 @@ export default (options={}) => {
 				// transform model state to postgest queries
 				//queries: [...ambient_queries, ...paging_queries]
 			}).then(response => {
-					// gather begin/end/count
-					let [_range, _count] = _xhr.getResponseHeader('content-range').split('/')
-					let [_begin, _end] = _range.split('-').map(v => ~~v)
+				// gather begin/end/count
+				let [_range, _count] = _xhr.getResponseHeader('content-range').split('/')
+				let [_begin, _end] = _range.split('-').map(v => ~~v)
+
+				// update count if presented
+				if (_count != '*') _cache.count = _count
 
-					// update count if presented
-					if (_count != '*') _cache.count = _count
+				// see if an upstream limit is exposed
+				if (_end - _begin + 1 < limit) _cache.upstream_limit = _end - _begin + 1
 
-					// fill the data cache
-					response.forEach((data, i) => {
-						//					_value_filters.forEach(([label, config]) => data[config.alias || label] = config.value_filter(data[config.alias || label]))
-						_cache.data[_begin + i] = data
-					})
+				// fill the data cache
+				response.forEach((data, i) => {
+					//					_value_filters.forEach(([label, config]) => data[config.alias || label] = config.value_filter(data[config.alias || label]))
+					_cache.data[_begin + i] = data
+				})
 
-					// for promise
+				// assert offset/limit and returned range
+				if (offset != _begin || _end - _begin + 1 > limit)
+					throw 'The request and response data range mismatches!'
+				if (_end - _begin + 1 < limit)
+					console.warn('The response range is narrower than requested, probably due to an upstream hard limit.')
+
+				// return
 				return _cache.data.slice(_begin, _end + 1)
 			})
 		}
@@ -100,10 +120,7 @@ export const default1 = (args) => {
 
 	return {
 		// model & model state
-		list: [],
-		count: null,
 		fully_loaded: false,
-		xhr: null,
 
 		// ambient
 		ambient: null,
@@ -134,14 +151,8 @@ export const default1 = (args) => {
 
 		// load data
 		load(args) {
-			args = args || {begin: 0}
-
 			let ambient_queries = this.reset()
 
-			// be lazy: if the data is presented, return the value immediately
-			if (this.list.slice(args.begin, args.end).length == args.limit && this.list.slice(args.begin, args.end).every(data => data != undefined))
-				return Promise.resolve(this.list.slice(args.begin, args.end))
-
 			// create limit&offset for postgrest
 			let paging_queries = []
 			if (args.begin != undefined) {