gptforfree commited on
Commit
823f1fb
·
verified ·
1 Parent(s): dfccb40

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +205 -3
README.md CHANGED
@@ -1,3 +1,205 @@
1
- ---
2
- license: apache-2.0
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # OpenChatData
2
+
3
+ OpenChatData is an anonymized dataset derived from database dumps from a discontinued AI chatbot service that routed model requests through OpenRouter.
4
+
5
+ The dataset contains **20,949 chat-log records** collected between **February 4, 2026 and April 5, 2026**, covering usage across **27 model identifiers**.
6
+
7
+ > **Important:** OpenChatData does **not** contain the text of user prompts or model responses. The released data consists of metadata and aggregate measurements such as token, word, and character counts.
8
+
9
+ ## Dataset Overview
10
+
11
+ Each row represents a recorded chatbot interaction/request.
12
+
13
+ The dataset includes:
14
+
15
+ * Date of the interaction
16
+ * Coarse geographic information
17
+ * OpenRouter model identifier
18
+ * Prompt/query token counts
19
+ * Model response token counts
20
+ * Word counts
21
+ * Character counts
22
+ * Whether the record was marked hidden by the original service
23
+
24
+ The supplied dataset contains **15 columns** and **20,949 rows**.
25
+
26
+ ## Schema
27
+
28
+ | Column | Description |
29
+ | ----------------------- | ----------------------------------------------------------------------------------------- |
30
+ | `date` | Date associated with the interaction |
31
+ | `city` | City associated with the request |
32
+ | `state` | State/region associated with the request |
33
+ | `country` | Country code associated with the request |
34
+ | `openrouter_model_slug` | Model identifier used for the request |
35
+ | `query_tokens` | Number of tokens in the user's query |
36
+ | `response_tokens` | Number of tokens in the model response |
37
+ | `total_tokens` | Combined query and response token count |
38
+ | `query_words` | Number of words in the user's query |
39
+ | `response_words` | Number of words in the model response |
40
+ | `total_words` | Combined query and response word count |
41
+ | `query_chars` | Number of characters in the user's query |
42
+ | `response_chars` | Number of characters in the model response |
43
+ | `total_chars` | Combined query and response character count |
44
+ | `is_hidden` | Integer flag indicating whether the interaction was marked hidden in the original service |
45
+
46
+ ## Basic Statistics
47
+
48
+ | Property | Value |
49
+ | ------------------------ | ----------------------: |
50
+ | Records | 20,949 |
51
+ | Columns | 15 |
52
+ | Date range | 2026-02-04 – 2026-04-05 |
53
+ | Unique model identifiers | 27 |
54
+ | Unique countries | 5 |
55
+ | Unique cities | 5 |
56
+ | Records marked hidden | 18,684 |
57
+
58
+ No missing values are present in the 15 released columns.
59
+
60
+ ## Models
61
+
62
+ Requests in the dataset were routed through a variety of models available through OpenRouter. Model names are preserved as they appeared in the original database.
63
+
64
+ As a result, model identifiers are **not necessarily normalized**. For example, identifiers such as:
65
+
66
+ ```text
67
+ gpt-4.1
68
+ openai/gpt-4.1
69
+ ```
70
+
71
+ may refer to the same underlying model while appearing as separate values in the dataset.
72
+
73
+ Some of the most frequently occurring identifiers include:
74
+
75
+ | Model identifier | Records |
76
+ | --------------------------------- | ------: |
77
+ | `gpt-4.1` | 4,717 |
78
+ | `openai/gpt-4o` | 3,414 |
79
+ | `openai/gpt-3.5-turbo` | 2,472 |
80
+ | `openai/gpt-4.1` | 2,152 |
81
+ | `openai/gpt-oss-120b` | 2,007 |
82
+ | `openai/gpt-oss-120b:free` | 1,268 |
83
+ | `qwen/qwen-max` | 1,068 |
84
+ | `nvidia/nemotron-nano-9b-v2:free` | 1,001 |
85
+
86
+ Researchers may wish to normalize model identifiers before performing model-level comparisons.
87
+
88
+ ## Privacy and Anonymization
89
+
90
+ OpenChatData originates from database dumps belonging to an abandoned chatbot service.
91
+
92
+ The public dataset intentionally excludes the actual contents of conversations. In particular, it does not include:
93
+
94
+ * User prompts
95
+ * Model response text
96
+ * Usernames or display names
97
+ * Email addresses
98
+ * Passwords or authentication credentials
99
+ * IP addresses
100
+ * Session or conversation contents
101
+
102
+ The remaining geographic fields should still be treated as potentially sensitive metadata. Users of the dataset should avoid attempts to identify, profile, or deanonymize individuals represented by the data.
103
+
104
+ The `is_hidden` field is retained as metadata from the original service. A hidden record should **not** be interpreted as permission to reconstruct, infer, or expose the underlying conversation.
105
+
106
+ ## Potential Uses
107
+
108
+ OpenChatData may be useful for research and analysis involving:
109
+
110
+ * LLM usage patterns
111
+ * Model popularity
112
+ * Prompt and response length distributions
113
+ * Token consumption
114
+ * Differences in response length between models
115
+ * Usage patterns over time
116
+ * Model-routing analysis
117
+ * OpenRouter ecosystem analysis
118
+ * Synthetic workload generation
119
+ * LLM infrastructure and capacity-planning research
120
+
121
+ Because conversation text is not included, the dataset is better suited to **usage and workload analysis** than semantic analysis of chatbot conversations.
122
+
123
+ ## Example
124
+
125
+ ```python
126
+ import pandas as pd
127
+
128
+ df = pd.read_csv("openrouter_chat_data.csv")
129
+
130
+ print(df.shape)
131
+ print(df.columns)
132
+
133
+ # Most frequently used models
134
+ print(df["openrouter_model_slug"].value_counts().head(10))
135
+
136
+ # Average token usage by model
137
+ model_stats = (
138
+ df.groupby("openrouter_model_slug")["total_tokens"]
139
+ .agg(["count", "mean", "median"])
140
+ .sort_values("count", ascending=False)
141
+ )
142
+
143
+ print(model_stats)
144
+ ```
145
+
146
+ ## Data Provenance
147
+
148
+ The data was recovered from database dumps from a now-discontinued AI chatbot service.
149
+
150
+ The service used OpenRouter to provide access to multiple language models. OpenChatData is an archival release of anonymized metadata extracted from those records rather than an official OpenRouter dataset.
151
+
152
+ **OpenChatData is not affiliated with, endorsed by, or produced by OpenRouter or the model providers represented in the dataset.**
153
+
154
+ Model identifiers and provider names remain in the dataset because they describe which models handled requests at the time the service was operating.
155
+
156
+ ## Limitations
157
+
158
+ There are several important limitations when interpreting this dataset.
159
+
160
+ **Sampling bias.** The records represent users of one chatbot service and should not be treated as representative of OpenRouter traffic, global LLM usage, or the general population.
161
+
162
+ **Model-name inconsistencies.** Model identifiers were stored by the original application and may contain aliases, prefixes, or naming conventions that changed over time.
163
+
164
+ **Geographic precision.** Location fields reflect information stored by the original service and may be inaccurate, approximate, VPN-influenced, or otherwise unrelated to a user's actual location.
165
+
166
+ **No conversation semantics.** Since prompt and response contents have been removed, the dataset cannot directly answer questions about conversation topics, intent, response quality, safety, or sentiment.
167
+
168
+ **Historical snapshot.** The dataset reflects a limited period of activity from February through April 2026 and should not be interpreted as describing current model availability or popularity.
169
+
170
+ **Application-specific behavior.** Token counts, hidden flags, routing decisions, and other fields may reflect implementation details of the original chatbot service.
171
+
172
+ ## Responsible Use
173
+
174
+ This dataset is intended for research, education, statistical analysis, and other legitimate uses.
175
+
176
+ Do not use OpenChatData to attempt to identify or deanonymize former users of the service. Researchers publishing analyses based on the dataset should consider aggregating geographic and low-frequency categories where necessary to avoid exposing potentially identifying usage patterns.
177
+
178
+ ## Citation
179
+
180
+ If you use OpenChatData in research, please cite the dataset and its repository.
181
+
182
+ A formal citation can be added here once a DOI or permanent repository URL is available.
183
+
184
+ ```bibtex
185
+ @dataset{openchatdata2026,
186
+ title = {OpenChatData},
187
+ year = {2026},
188
+ note = {An anonymized dataset of OpenRouter-backed chatbot usage metadata}
189
+ }
190
+ ```
191
+
192
+ ## License
193
+
194
+ OpenChatData is licensed under the **Apache License 2.0**.
195
+
196
+ You may use, modify, and redistribute the dataset in accordance with the terms of the Apache License, Version 2.0. See the repository's [`LICENSE`](LICENSE) file for the full license text.
197
+
198
+ Unless required by applicable law or agreed to in writing, the dataset is distributed on an **"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND**, either express or implied.
199
+
200
+
201
+ ## Disclaimer
202
+
203
+ OpenChatData is an independent archival dataset derived from a discontinued third-party chatbot service.
204
+
205
+ OpenRouter and the individual model providers represented by model identifiers in the dataset are not responsible for the collection, preparation, anonymization, publication, or maintenance of this dataset.