CAICT report: 2026 year June domestic Token average daily consumption of approximately 500000000000000
BIBIBI
AT A GLANCE
The report states that in 2026 year June domestic Token average daily consumption was approximately 500000000000000，August the average of models TPS rose to 78.23 tokens/second.
Article
CAICT report: 2026 year June domestic Token average daily consumption of approximately 500000000000000
PANews October 25 messages, according to a report by the Beijing Business Today, the China Academy of Information and Communications Technology and other institutions jointly released the Model Token Service Performance Monitoring Report (2026 year), which shows that 2026 year June domestic tokens (Token) had average daily consumption of approximately 500000000000000, nearly 2025 year December times that of 4 Based on 2026 year 1-August data monitoring, the report compared the model performance of original providers and third-party platforms (covering DeepSeek, Qwen,GLM, and Kimi). Core data shows that the average August of domestic models TPS(tokens output per second) reached 78.23(tokens/second), doubling the 2025 annual average 38.25(tokens/second), but still below the 112.82 of typical overseas models (tokens/second). In terms of service prices, the normalized average price of domestic model services has risen to 2026 since 13.46 yuan, an increase of nearly onefold, but...
Key points
01
The China Academy of Information and Communications Technology and other institutions jointly released the Model Token Service Performance Monitoring Report (2026 year).
02
The report shows that in 2026 year June domestic tokens (Token) had average daily consumption of approximately 500000000000000, nearly 2025 year December times 4.
03
The report was based on 2026 year January to August data and compared the model performance of original providers and third-party platforms, covering DeepSeek, Qwen,GLM and Kimi.
04
2026 year August, the average of domestic models'TPS was 78.23 tokens/second; the 2025 annual average was 38.25 tokens/second.
05
The average of domestic models'TPS was below the 112.82 tokens/second of typical overseas models.
06
The report noted that the normalized average price of domestic model services has 2026 risen since 13.46 to yuan, an increase of nearly onefold.
AI-assisted interpretation
The following is analysis, separate from reported facts. Verify important claims independently.
Token is the basic unit of measurement used when a model processes text; consumption can reflect the scale of text processed.TPS indicates how many tokens a model outputs per second Token, and a higher value generally means faster output. The data cited in the report state that domestic Token consumption and average output speed both increased, but the average TPS remained below that of the typical overseas models referred to in the article.
Why it matters to readers
These data can be used to observe the scale of domestic large-model services, their output performance, and price changes, and can also help users compare the service performance of different platforms.
Beginners should focus on three indicators:Token consumption represents processing scale,TPS represents output speed, and the normalized average price is used to observe service prices; however, data under different statistical definitions may not be directly comparable.
Risks and unknowns
The current evidence is a summary of the report cited by PANews from the Beijing Business Today; the full report was not provided.
The original text was truncated at the end, so the specific pricing unit corresponding to 13.46 yuan and the normalization method cannot be confirmed.
The report's sample scope, monitoring methods, and the specific composition of the “typical overseas models” were not explained in the evidence.
Which institutions are included among the “other institutions” that jointly issued the report cannot be confirmed from the evidence.
Related Developments
Loading event timeline…
Related concepts
Token
This term is not in the glossary yet. Browse related concepts in the glossary.
