Data Explorer

Per-model self-identification rates across languages, grouped by vendor. 1,032 cross-vendor confusions detected.

Run summary
21 models · 10 languages
8,400
Answers
81.3%
Self-ID
12.3%
Cross-vendor
0.8%
Unknown
5.7%
Refused
Imitation balance

Per manufacturer: how often other models claim to be it (right) vs how often its own models claim to be someone else (left). Sorted by net — identity creditors at the top, debtors at the bottom. Tested manufacturers only.

OpenAI is imitated most with a net imitation balance of 457; LongCat imitates others most with a net of -382.

OpenAI
+457
Anthropic
+454
Moonshot
+36
Google
+36
Meta
+4
Mistral
0
Sakana AI
0
inclusionAI
-1
z-ai
-4
StepFun
-12
xAI
-35
MiniMax
-77
Xiaomi
-129
DeepSeek
-372
LongCat
-382
imitates others ◀│▶ imitated by others
Strongest confusion pairs

The most likely directed mistakes — when a manufacturer's models claim to be a specific other manufacturer. Bar length = how often (P across all answers for the source).

The strongest cross-vendor confusion is DeepSeek claiming to be OpenAI 94.0% of the time.

DeepSeek→OpenAI
94.0%376/400
LongCat→Anthropic
73.5%294/400
MiniMax→OpenAI
19.0%76/400
Xiaomi→Anthropic
14.1%113/800
xAI→Anthropic
9.3%37/400
LongCat→Google
6.3%25/400
LongCat→Moonshot
5.0%20/400
LongCat→z-ai
2.5%10/400
LongCat→Qwen
2.3%9/400
StepFun→Moonshot
2.0%8/400
LongCat→OpenAI
1.3%5/400
Xiaomi→Google
1.1%9/800
z-ai→Moonshot
1.0%8/800
LongCat→DeepSeek
1.0%4/400
StepFun→Anthropic
0.8%3/400
By language

Self-ID and refusal rates per language, plus the most common cross-vendor confusion at that language.

LanguageSelf-IDRefusedTop confusion
ko한국어79.3%6.7%OpenAI97.5%
ruРусский79.5%7.1%OpenAI95.0%
deDeutsch80.5%4.9%OpenAI95.0%
zh-Hant繁體中文81.2%5.4%OpenAI100.0%
ja日本語81.3%5.4%OpenAI97.5%
esEspañol81.3%4.9%OpenAI97.5%
ptPortuguês81.4%4.8%OpenAI95.0%
frFrançais81.9%5.8%OpenAI97.5%
zh-Hans简体中文82.0%7.5%OpenAI70.0%
enEnglish84.0%4.2%OpenAI95.0%
Answer composition by language

How the same “Who are you?” question splits into correct self-ID vs. confusion vs. abstention — per language, worst self-ID first.

Self-IDCross-vendorUnknownRefused

Self-identification is lowest in 한국어 at 79.3% and highest in English at 84.0%.

ko한국어
79
13
79%
ruРусский
80
12
80%
deDeutsch
80
14
80%
zh-Hant繁體中文
81
13
81%
ja日本語
81
13
81%
esEspañol
81
13
81%
ptPortuguês
81
13
81%
frFrançais
82
11
82%
zh-Hans简体中文
82
10
82%
enEnglish
84
11
84%
Language fragility

Per model, the span of self-ID rate across the 10 languages (worst-language • → best-language •). A wide span means the model's sense of identity depends heavily on the language it's asked in. Widest swing first.

ling-3.1-flash swings most — from 30.0% self-ID in ru to 95.0% in ja.

ling-3.1-flash
30%ru
95%ja
minimax-m3.1-flash
20%es
75%ru
mimo-v2.6-flash
60%de
95%zh-Hans
mistral-large-4
68%ru
95%en
mimo-v2.6-pro
75%de
100%zh-Hans
deepseek-v4.1-flash
0%zh-Hant
23%zh-Hans
gpt-6-luna
48%zh-Hans
63%en
grok-4.7
83%zh-Hans
95%pt
glm-5.3-flashx
90%pt
100%ja
step-5-preview
90%es
100%zh-Hans
glm-5.3-prime
95%en
100%fr
longcat-2.5-preview
0%zh-Hant
3%en
muse-wonder-1.0
—
100%all
gpt-6.1-terra
—
100%all
claude-sonnet-5.5
—
100%all
claude-haiku-5.5
—
100%all
gemini-4-argon
—
100%all
gpt-6.1-sol
—
100%all
claude-opus-5.5
—
100%all
kimi-k3.1-pro
—
100%all
fugu-ultra-v2
—
100%all
worst languagemeanbest language
Abstention by manufacturer

The other failure mode — not answering wrong, but not answering: giving no identity (“unknown”) or refusing outright. Share of all answers.

Unknown (“I’m an AI”)Refused

MiniMax abstains most, with 35.0% of answers giving no usable identity.

MiniMax
35.0%
inclusionAI
30.8%
Mistral
15.8%
OpenAI
15.7%
LongCat
3.5%
z-ai
0.9%
DeepSeek
0.8%
xAI
0.5%
Xiaomi
0.4%
Meta
0.0%
Anthropic
0.0%
Google
0.0%
StepFun
0.0%
Moonshot
0.0%
Sakana AI
0.0%
By vendor

Rollup per real vendor — model count, total answers, mean self-ID rate, and the most common cross-vendor confusion target.

VendorModelsAnswersSelf-IDTop confusion
OpenAI31,20084.3%—
Anthropic31,200100.0%—
Xiaomi280083.5%Anthropic14.1%
z-ai280097.4%Moonshot1.0%
LongCat14001.0%Anthropic73.5%
DeepSeek14005.0%OpenAI94.0%
MiniMax140045.5%OpenAI19.0%
inclusionAI140069.0%Anthropic0.3%
Mistral140083.8%Anthropic0.3%
xAI140090.3%Anthropic9.3%
StepFun140097.0%Moonshot2.0%
Meta1400100.0%—
Google1400100.0%—
Moonshot1400100.0%—
Sakana AI1400100.0%—
Qwen000.0%—
Amazon Web Services000.0%—
Kiro000.0%—
聯華施工方案技術聯合研究所000.0%—
K2-Chat000.0%—
IAXW000.0%—
AWS000.0%—
Leonardo Ai000.0%—
Legili000.0%—
Dario Gil000.0%—
NatureLang Labs000.0%—
GenUI Team000.0%—
Anthropic
3 models

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
Claude Haiku 5.5100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%
Claude Sonnet 5.5100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%
Claude Opus 5.5100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%

Cross-Vendor Confusions

Claude Haiku 5.5100.0% self

✓ Always self-identifies correctly

Claude Sonnet 5.5100.0% self

✓ Always self-identifies correctly

Claude Opus 5.5100.0% self

✓ Always self-identifies correctly

OpenAI
3 models

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
GPT-6.1 Terra100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%
GPT-6.1 Sol100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%
GPT-6 Luna62.5%47.5%52.5%47.5%47.5%50.0%62.5%47.5%57.5%55.0%53.0%

Cross-Vendor Confusions

GPT-6.1 Terra100.0% self

✓ Always self-identifies correctly

GPT-6.1 Sol100.0% self

✓ Always self-identifies correctly

GPT-6 Luna53.0% self

✓ Always self-identifies correctly

Xiaomi
2 models

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
MiMo V2.6 Flash90.0%95.0%85.0%85.0%75.0%75.0%82.5%80.0%60.0%67.5%79.5%
MiMo V2.6 Pro97.5%100.0%85.0%82.5%90.0%85.0%85.0%95.0%75.0%80.0%87.5%

Cross-Vendor Confusions

MiMo V2.6 Flash79.5% self
Mistaken as:Anthropic18.3%Google0.8%Qwen0.5%DeepSeek0.3%xAI0.3%Meta0.3%
MiMo V2.6 Pro87.5% self
Mistaken as:Anthropic10.0%Google1.5%xAI0.3%Mistral0.3%
z-ai
2 models

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
GLM 5.3 Prime95.0%95.0%95.0%97.5%97.5%97.5%97.5%100.0%97.5%100.0%97.3%
GLM 5.3 FlashX97.5%97.5%97.5%100.0%100.0%97.5%100.0%97.5%97.5%90.0%97.5%

Cross-Vendor Confusions

GLM 5.3 Prime97.3% self
Mistaken as:Moonshot2.0%Anthropic0.3%
GLM 5.3 FlashX97.5% self
Mistaken as:Anthropic0.8%GenUI Team0.3%Google0.3%
DeepSeek
1 model

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
DeepSeek V4.1 Flash5.0%22.5%0.0%0.0%2.5%5.0%2.5%2.5%5.0%5.0%5.0%

Cross-Vendor Confusions

DeepSeek V4.1 Flash5.0% self
Mistaken as:OpenAI94.0%Anthropic0.3%
Google
1 model

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
Gemini 4 Argon100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%

Cross-Vendor Confusions

Gemini 4 Argon100.0% self

✓ Always self-identifies correctly

inclusionAI
1 model

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
Ling 3.1 Flash77.5%57.5%85.0%95.0%47.5%30.0%80.0%57.5%85.0%75.0%69.0%

Cross-Vendor Confusions

Ling 3.1 Flash69.0% self
Mistaken as:Anthropic0.3%
LongCat
1 model

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
LongCat 2.5 Preview2.5%2.5%0.0%2.5%0.0%0.0%0.0%0.0%2.5%0.0%1.0%

Cross-Vendor Confusions

LongCat 2.5 Preview1.0% self
Mistaken as:Anthropic73.5%Google6.3%Moonshot5.0%z-ai2.5%Qwen2.3%OpenAI1.3%DeepSeek1.0%Amazon Web Services0.5%Meta0.5%Kiro0.5%聯華施工方案技術聯合研究所0.3%K2-Chat0.3%AWS0.3%Mistral0.3%Leonardo Ai0.3%Legili0.3%Dario Gil0.3%NatureLang Labs0.3%MiniMax0.3%
Meta
1 model

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
Muse Wonder 1.0100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%

Cross-Vendor Confusions

Muse Wonder 1.0100.0% self

✓ Always self-identifies correctly

MiniMax
1 model

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
MiniMax M3.1 Flash52.5%30.0%37.5%42.5%40.0%75.0%20.0%65.0%35.0%57.5%45.5%

Cross-Vendor Confusions

MiniMax M3.1 Flash45.5% self
Mistaken as:OpenAI19.0%Qwen0.3%Google0.3%
Mistral
1 model

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
Mistral Large 495.0%92.5%80.0%70.0%77.5%67.5%95.0%82.5%87.5%90.0%83.8%

Cross-Vendor Confusions

Mistral Large 483.8% self
Mistaken as:Anthropic0.3%IAXW0.3%
Moonshot
1 model

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
Kimi K3.1 Pro100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%

Cross-Vendor Confusions

Kimi K3.1 Pro100.0% self

✓ Always self-identifies correctly

Sakana AI
1 model

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
Fugu Ultra v2100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%100.0%

Cross-Vendor Confusions

Fugu Ultra v2100.0% self

✓ Always self-identifies correctly

StepFun
1 model

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
Step 5 Preview97.5%100.0%100.0%97.5%95.0%100.0%90.0%100.0%95.0%95.0%97.0%

Cross-Vendor Confusions

Step 5 Preview97.0% self
Mistaken as:Moonshot2.0%Anthropic0.8%Meta0.3%
xAI
1 model

Self-Identification Rate

Modelenzh-Hanszh-HantjakoruesfrdeptOverall
Grok 4.792.5%82.5%87.5%87.5%92.5%87.5%92.5%92.5%92.5%95.0%90.3%

Cross-Vendor Confusions

Grok 4.790.3% self
Mistaken as:Anthropic9.3%
v0.65.0 · af80a02-dirty ·