Overview›Pagination

APIFlow-Bench 1.0 Leaderboard | Pagination

Whether a model walks paginated collections to completion — cursors, offsets, and page limits — instead of answering from the first page. 51 tasks × 24 models in this slice.

Ranking◴ Generated 2026-07-31☑ 6,120 trials▤ 24 models

Pass rate

82.9%

Error Rate

0.1%

Trials

6,120

Tasks

Models

p95 Time

15.5m

p95 Tokens

1,111,810

Tool Calls Avg

24.1

Rank	Rank Spread	Model	Pass rate ± 90% CI	Pass (no abstain)	Abstain%	p95 Time	p95 Tokens	Avg Tools
1	1 ↔ 11	gpt-5.5	92.2% 86.3%–98.0%	94.0% 235/250	2.0% 5/255	2.8m	411,784	19.3
2	1 ↔ 12	gpt-5.6-sol	91.8% 85.5%–97.6%	94.0% 234/249	2.4% 6/255	11.5m	332,659	17.5
3	1 ↔ 13	deepseek-v4-flash	91.4% 85.1%–96.9%	92.1% 233/253	0.8% 2/255	3.7m	883,444	25.6
3	1 ↔ 14	kimi-k2p7-code	91.4% 84.7%–97.3%	92.5% 233/252	1.2% 3/255	13.0m	678,749	17.9
5	1 ↔ 14	claude-haiku-4-5-20251001	91.0% 84.3%–96.9%	91.7% 232/253	0.8% 2/255	2.0m	585,238	20.1
5	1 ↔ 14	claude-opus-4-8	91.0% 84.7%–96.5%	94.7% 232/245	3.9% 10/255	2.8m	277,962	14.2
5	1 ↔ 14	claude-sonnet-5	91.0% 84.3%–96.5%	92.8% 232/250	2.0% 5/255	4.6m	630,162	19.2
5	1 ↔ 14	claude-sonnet-4-6	91.0% 85.1%–96.5%	91.0% 232/255	0.0% 0/255	9.8m	964,534	30.1
9	2 ↔ 15	kimi-k3	90.6% 83.5%–96.5%	90.6% 231/255	0.0% 0/255	38.8m	1,239,877	26.7
10	2 ↔ 16	gpt-5.6-terra	90.2% 83.5%–96.1%	92.4% 230/249	2.4% 6/255	3.8m	158,636	14.4
10	2 ↔ 16	kimi-k2p6	90.2% 82.4%–96.1%	90.2% 230/255	0.0% 0/255	11.1m	2,214,243	22.9
12	3 ↔ 17	kimi-k2p5	89.4% 82.4%–96.1%	90.8% 228/251	1.6% 4/255	2.7m	457,528	17.3
12	3 ↔ 16	glm-5p2	89.4% 82.7%–95.3%	89.4% 228/255	0.0% 0/255	30.2m	1,848,907	21.6
14	5 ↔ 18	gpt-5.6-luna	88.6% 81.6%–94.9%	90.0% 226/251	1.6% 4/255	4.3m	152,372	14.8
15	7 ↔ 20	deepseek-v4-pro	87.5% 80.0%–94.1%	87.8% 223/254	0.4% 1/255	18.1m	824,320	23.9
16	11 ↔ 21	minimax-m2p7	86.3% 78.4%–92.9%	87.6% 220/251	1.6% 4/255	1.8m	324,539	15.1
16	11 ↔ 21	qwen3p6-plus	86.3% 79.6%–92.9%	86.6% 220/254	0.4% 1/255	11.7m	1,215,543	23.7
18	14 ↔ 21	gpt-5-mini	84.7% 77.6%–91.4%	90.4% 216/239	6.3% 16/255	1.4m	160,086	12.7
18	13 ↔ 21	gpt-oss-20b	84.7% 77.6%–91.0%	87.1% 216/248	2.7% 7/255	4.2m	159,641	16
18	14 ↔ 21	qwen3p7-plus	84.7% 77.6%–91.4%	86.1% 216/251	1.6% 4/255	11.7m	554,673	14
21	14 ↔ 21	gpt-oss-120b	84.3% 76.5%–91.4%	84.6% 215/254	0.4% 1/255	49.7s	164,607	15.1
22	22 ↔ 22	gpt-5.4-mini	70.2% 62.0%–78.4%	75.5% 179/237	7.1% 18/255	24.0s	42,407	11.4
23	23 ↔ 23	minimax-m3	34.1% 27.4%–41.6%	58.4% 87/149	41.6% 106/255	66.1m	15,061,100	159.3
24	24 ↔ 24	claude-fable-5	16.9% 9.4%–24.7%	16.9% 43/255	0.0% 0/255	3.9m	66,685	5.7

Diagnostics

Loading the full trial dataset for this view…