Benchmarks

Rank the signal, inspect the evidence.

Blind RouterPlex preferences and dated external snapshots, shown beside latency, evaluation cost, sample size, and confidence.

Ranking console

Published leaderboard.

Select a task and scoring signal. Every row keeps the underlying evidence visible.

overallcodingreasoningwritinggeneral
Quality signalExternal signal
Rank / modelQuality signalArenaExternalLatencyEval costVotesEvidence
01
99.7
-99.7--0Cited
02
99.2
-99.2--0Cited
03
98.5
-98.5--0Cited
04
95.5
-95.5--0Cited
05
95.2
-95.2--0Cited
06
95.0
-95.0--0Cited
07
GPT-5.5

OpenAI

94.5
-94.5--0Cited
08
93.5
-93.5--0Cited
09
92.5
-92.5--0Cited
10
91.7
-91.7--0Cited
11
GPT-5.4

OpenAI

89.9
-89.9--0Cited
12
GLM-5.1

Zhipu

89.7
-89.7--0Cited
13
89.2
-89.2--0Cited
14
Kimi K2.6

Moonshot

88.2
-88.2--0Cited
15
87.9
-87.9--0Cited
16
85.7
-85.7--0Cited
17
Hy3

Tencent Hunyuan

85.2
-85.2--0Cited
18
82.4
-82.4--0Cited
19
MiniMax M3

MiniMax

82.2
-82.2--0Cited
20
MiMo V2.5

Xiaomi

80.4
-80.4--0Cited
21
79.4
-79.4--0Cited
22
70.9
-70.9--0Cited
23
67.6
-67.6--0Cited
0-100 normalized display scalegeneral / combined quality

External snapshot

Source and retrieval date stay attached to every imported signal.

LMArena Text Leaderboard2026-09-02 text_style_control b005f0dd4ad8 / retrieved 2026-09-10
Published leaderboard name2026-07-22 / retrieved 2026-07-22

Method / ranking-v1

Transparent by construction.

Normalize, never obscure

External results retain their raw rating, source rank, interval, vote count, version, and retrieval date. Source rank becomes a within-category percentile on the common 0-100 scale.

Confidence earns its weight

Blind Arena votes grow in influence only as valid sample confidence improves. Private prompts stay private and never enter the public ranking.

Run a blind comparison