Biteye AI Daily July 14

Today’s zone market capitalization: $10.28B +1.91%

The weekend saw little fluctuation, with continuous developments in decentralized AI and mainstream AI, showing sustained growth.

📈24h Coingecko Trading Volume Top 10: Token Trading Volume Market Capitalization Price Volatility 1- $BANANAS31 $407.1M $85.0M $0.01 -42.1% 2- $VIRTUAL $211.7M $1.1B $1.71 +0.3% 3- $TAO $109.7M $3.7B $396.23 +3.5% 4- $AI16Z $67.6M $204.6M $0.19 +3.4% 5- $AIXBT $62.6M $140.4M $0.15 +0.8% 6- $PROMPT $47.2M $38.0M $0.17 +17.1% 7- $GOAT $36.9M $125.5M $0.13 -0.6% 8- $SAHARA $36.4M $147.3M $0.07 +0.9% 9- $KAITO $36.3M $362.3M $1.50 -2.6% 10- $COOKIE $24.3M $102.4M $0.19 +1.3%

⭐️Key Information

1. @SaharaLabsAI DSP: Public beta starts on July 22

• DSP ——Data Services Platform, Data Services Platform
• Public testing: No barriers globally, everyone can earn data money.
• Participate to earn real tokens $SAHARA, and more ways to make money will be released later.
• Existing whitelist users can now participate in the MIA dual reward task: earn $SAHARA + $MIA at once (Entrance 👉 app.saharaai.com)

Comment: The world has long suffered from white list waiting, no need to苦等白名单 anymore, it is finally open to everyone!

2. @TargonCompute: Decentralized cloud rises by 10 billion tokens in a day

• As the 4th subnet of Bittensor, it focuses on high-performance inference, model leasing, and GPU services.
• The daily new call volume has soared to the hundred billion level within two weeks, with two-thirds being genuine enterprise payments.

Comment: The price is competitive, performance is also solid, and decentralized computing power is starting to emerge.

3. @Kimi_Moonshot released Kimi K2: domestically produced 10 trillion parameter MoE bomb.

• 1 T total parameters / 32B active experts, outstanding performance in programming and agent-type tasks.
• In the three major public evaluations, it ranks first among open source models: SWE Bench (code modification), Tau2 (logical reasoning), AceBench (blockchain tasks).
• Multi-modal/“thinking mode” is not supported at this time.
• MoE is a special neural network architecture that requires complex gating mechanisms and expert optimization, resulting in high training costs. However, the inference process can save computational resources and is suitable for multiple scenarios.

Comment: Domestic models still perform well among open-source large models, and major computing power platforms will soon be integrated.

4. Goldman Sachs trials @DevinAI: AI engineers are really on the job

• Devin is built by @cognition_labs, allowing users to write code, run tests, and change languages.
• Goldman Sachs first lets it do the repetitive tasks that “engineers least want to do”, under human supervision.
• Cognition Labs was established in 2023, with a valuation of nearly 4 billion dollars.

Comment: The application of AI has quietly moved from Silicon Valley to Wall Street, with a larger scale, more practical applications, and greater attention.

📊 Data source: @xhunt_ai, Cookie DAO, Coingecko

💡Risk Warning: The above is for information sharing only and not investment advice.