Daily AI RoundupsWednesday, 24 June 2026

Hardware and money led 24 June. OpenAI unveiled Jalapeño, its first AI chip built with Broadcom for ChatGPT, Codex and API workloads, Qualcomm agreed to acquire Modular to expand its hardware-independent compute platform, and Menlo Ventures raised a $3B fund, its largest in 50 years. NVIDIA pushed inference with DFlash, claiming up to 15x faster Blackwell decoding, and made cuVS vector indexing default in OpenSearch Serverless. On models and agents, Qwen-AgentWorld-397B topped AgentWorldBench at 58.71, Mistral opened OCR 4 via API, Claude Code 2.1.187 shipped 21 CLI changes, Nous Research's Hermes Agent added /learn, and Unitree advertised its R1 humanoid from $4,900.

• AI chips
• AI deals
• AI funding
• Model releases
• Robotics

Stories in this roundup

1. 01
   OpenAI unveils Jalapeño, its first AI chip built with Broadcom for ChatGPT, Codex and API workloadsOPENAI
2. 02
   Menlo Ventures raised $3B, its largest in 50 years, to back AI startups across sectorsCRUNCHBASE
3. 03
   Qwen-AgentWorld-397B-A17B scores 58.71 on 7-domain AgentWorldBench, topping GPT-5.4QWEN
4. 04
   NVIDIA says DFlash boosts Blackwell LLM inference up to 15x via lightweight draft modelNVIDIA
5. 05
   Qualcomm agrees to acquire Modular to expand hardware-independent AI compute platformMODULAR
6. 06
   Unitree advertises the R1 humanoid robot from $4,900 with ready stockUNITREE
7. 07
   Claude Code 2.1.187 ships 21 CLI changes with secret blocking and 5-minute remote MCP abortsCLAUDE CODE
8. 08
   Replit details ViBench, A/B tests and Telescope for improving coding agents at scaleREPLIT
9. 09
   Nous Research’s Hermes Agent adds /learn to turn code, PDFs and docs into reusable verifiable skillsNOUS RESEARCH
10. 10
    NVIDIA and AWS make cuVS vector indexing default in OpenSearch Serverless, add RTX PRO 4500 EC2 G7 for AINVIDIA
11. 11
    SpaceX launches $25B notes offering for debt repayment and AI expansionCHANNEL NEWS ASIA
12. 12
    Mistral AI makes OCR 4 available via API, Studio, SageMaker, Foundry and self-hostingMISTRAL
13. 13
    GPT-5.6 speculation says model could be costlier or larger than GPT-5.5, based on screenshotsSCALING01
14. 14
    Legal-tech firm Legion sued the US over an order limiting foreign access to top-tier Anthropic modelsREUTERS