AI News
AI community posts, news and release updates
Latest updates
Modal offers Kimi K3 inference API with $30/month free compute credits, covering ~10M input or 2M output K3 tokens. One-click OpenCode setup, compatible with OpenAI tools. Also covers other models and raw GPU. No surprise bills, open to everyone.
- free API
- free compute
- Kimi K3
- Modal
- $30 credits
Reports reveal that Gemini is on the list of tools banned for Google engineers, ironically a Google product itself, raising questions about internal AI tool safety and reliability.
- Gemini
- banned tools
- engineers
A user reports that reverting to the Fable model gave worse performance than Opus 4.6-8, caused major issues in Hermes Agent, and they will not use it again.
- Claude
- Anthropic
- Hermes Agent
- Fable
Nvidia has formed the Open Secure AI Alliance, but major AI players OpenAI, Google, and Anthropic are notably absent, raising questions about their commitment to collaborative AI safety.
- Anthropic
- AI safety
- OpenAI
- Nvidia
- Open Secure AI Alliance
- Launch
Microsoft Launches Its First Major Cybersecurity AI Model: MAI-Cyber-1-Flash
Published Jul 28, 2026Microsoft unveiled MAI-Cyber-1-Flash, a generative AI model for cybersecurity. Paired with GPT-5.4, it outperforms Anthropic's Mythos 5 and Google's 3.5 Flash Cyber on CyberGym benchmark.
- Microsoft
- cybersecurity
- AI model
- MAI-Cyber-1-Flash
- News
X is shutting down Grok Companions and Redditors are absolutely crashing out over it
Published Jul 28, 2026X is shutting down Grok Companions; Redditors are passionately upset about the decision.
- Grok
- controversy
- X
- shutdown
- Companions
- Rant
Anthropic Trained Claude on 500k Pirated Books, Settled for $1.5B, Then Lobbies to Block Chinese Labs from Using Its Outputs
Published Jul 27, 2026Anthropic downloaded 500,000 copyrighted books for training, got fair use ruling, paid $1.5B to settle piracy claims. Now CEO Dario Amodei calls Chinese labs' use of Claude outputs 'industrial-scale distillation' and pushes for policy to stop them, highlighting double standards.
- Claude
- Anthropic
- MiniMax
- Moonshot
- DeepSeek
- Distillation
- fair use
- Rant
GPT 5.6 and Claude get smarter but worse at human conversation, insufferably precise
Published Jul 27, 2026Rants that GPT 5.6 and Claude Opus/Fable 5 speak with insufferable precision, getting worse at talking to people as they get smarter, like world-class engineers.
- GPT-5.6
- model behavior
- precision
- Claude Fable 5
- Claude Opus 5
- Tutorial
[Breaking] Kimi K3 deployed on 8x NVIDIA B300 on Day 0: 2.8T MoE runs on single node without quantization
Published Jul 27, 2026On the same day weights were released, Kimi K3 (2.8T MoE) was deployed on a single node with 8x NVIDIA B300, running without extra quantization. Deployment method, GPU memory usage, inference and coding performance are reported.
- benchmark
- MoE
- deployment
- NVIDIA B300
- Kimi K3
Developer Boris Cherny advises deleting the Claude.md file every six months to prevent context pollution and maintain model performance.
- CLAUDE.md
- Boris Cherny
- tip
- cleanup