Wednesday, 8 July 2026
SpaceXAI ships Grok 4.5, built with Cursor
Archive item — written before sources were shown.
Grok 4.5 launched on a 1.5T-parameter base trained partly on real Cursor coding sessions, priced at $2/$6 per million tokens, well under Opus and GPT-5.6.
OpenAI launches GPT-Live voice model
Archive item — written before sources were shown.
GPT-Live-1 and GPT-Live-1 mini can listen and speak at once, backchanneling like a person, while handing hard queries to GPT-5.5 in the background.
OpenAI launches GPT-Live voice models
Archive item — written before sources were shown.
GPT-Live-1 and GPT-Live-1 mini replace Advanced Voice Mode in ChatGPT with full-duplex audio and real-time background task delegation to GPT-5.5.
SpaceXAI and Cursor ship Grok 4.5
Archive item — written before sources were shown.
Grok 4.5 is the first model built with Cursor since the $60B acquisition agreement, targeting enterprise work in legal, finance, and software engineering.
Bespoke Labs raises $40M to train agents
Archive item — written before sources were shown.
Bespoke Labs raised $40M (Wing VC) to build RL environments for training reliable AI agents, with angels from Anthropic, OpenAI, and Meta.
Meta's Muse Image launches free in Meta AI
Archive item — written before sources were shown.
Meta Superintelligence Labs released Muse Image, ranking second in human-preference image tests and free to use in Meta AI, Instagram Stories, and WhatsApp.
Norm AI hits unicorn with $120M round
Archive item — written before sources were shown.
Norm AI raised $120M Series C at a $1.2B valuation, building an AI-native law firm combined with supervisory AI for agents operating in regulated environments.
OpenAI starts charging for Workspace Agents
Archive item — written before sources were shown.
ChatGPT Workspace Agents in Business, Enterprise, and Edu plans began drawing workspace credits on July 6, ending the free preview that launched in April 2026.
SambaNova raises $1B at $11B valuation
Archive item — written before sources were shown.
SambaNova's Series F first close, led by General Atlantic, values the AI chipmaker at $11B and lands alongside a new JPMorganChase inference partnership.
Google overhauls its coding-agent benchmark
Archive item — written before sources were shown.
Android Bench switched to the Harbor evaluation framework and added eight models; Claude Fable 5 currently leads at 84.5 versus GPT-5.5's 80.2.
Prime Intellect raises $130M for agents
Archive item — written before sources were shown.
The $1B-valuation round, led by Radical Ventures, funds compute, an RL framework and eval tooling so enterprises can train their own agentic AI systems.
Claude scans 466M lines of code in 20 hours
Archive item — written before sources were shown.
Alberta deployed 50 Claude Code agents to scan 466M lines of code across 27 ministries in 20 hours — versus 6.5 years estimated manually.
Chinese models hit 30% of US OpenRouter use
Archive item — written before sources were shown.
Chinese AI models have held above 30% of US OpenRouter token share since February, peaking at 46%. Prices run 60-90% below US frontier models.
DeepSeek DSpark speeds inference 60-85%
Archive item — written before sources were shown.
DSpark cuts DeepSeek-V4-Flash latency 60-85% using semi-autoregressive generation and confidence-scheduled verification, deployed under live production traffic.
n8n 2.30 adds Microsoft Service Principal auth
Archive item — written before sources were shown.
n8n 2.30.0 adds Service Principal auth to Microsoft Excel, Teams, and To Do nodes, plus a unified webhook URL environment variable and 75+ bug fixes.
OpenAI confirms GPT-5.6 launch for July 9
Archive item — written before sources were shown.
GPT-5.6 Sol, Terra and Luna clear US Commerce Dept review and launch publicly Thursday; Sol prices at $5/$30 per million tokens, Luna at $1/$6.
OpenAI chief futurist Joshua Achiam departs
Archive item — written before sources were shown.
Achiam, who previously led OpenAI's now-disbanded mission-alignment team, departs after nearly nine years, joining recent AI-safety-linked exits.