AI News 2026-08-02
New To You · 12 of 346 stories

GPT 5.6 prices cut by up to 80% as OpenAI refutes major math conjecture.

50 sources swept 346 distinct stories from the last 3 days 70 already covered, held back 12 worth your time
  1. [AINews] GPT 5.6 price cut by 20%-80%: Cost of GPT 5.4 Intelligence dropped 13x in 4 months due to GPT 5.6 recursive self-optimization

    Cost of GPT 5.4 intelligence dropped 13x in four months due to recursive self-optimization.

    Latent Space··Models & releases·newsletter·latent.space
  2. AI keeps cracking unsolved math problems, and mathematicians have mixed feelings

    OpenAI published a counterexample to the Unit Distance Conjecture, a problem open since 1946.

    The Decoder··Research·press·the-decoder.com
  3. Google reveals Gemini Robotics 2.0, promising improved dexterity and safety

    Gemini Robotics 2 allows robots to continuously analyze changing environments and collaborate.

    Ars Technica AI··Models & releases·press·arstechnica.com
  4. Echoverse: Deep, evolving environments for computer-use agents

    Microsoft built twelve training worlds that reproduce real application behavior and maintain coherent state.

    Microsoft Research··Agents & tooling·lab·microsoft.com
  5. Ontologies Are So Back: Why AI Agents Are Reviving the Semantic Web

    Engineers are using ontologies to keep probabilistic agents inside deterministic boundaries.

    Latent Space··Agents & tooling·newsletter·latent.space
  6. Disrupting a Criminal Scam Operation

    OpenAI disrupted a Cambodia-based operation using ChatGPT for investment and romance scams.

    OpenAI··Safety & security·lab·openai.com
  7. Gemini Spark now integrates with Chrome

    Gemini Spark can now use logged-in accounts and saved passwords to handle web errands.

    Google (Gemini)··Products & apps·lab·blog.google
  8. datasette-apps 0.2a0

    Datasette Agent can now smoke test apps by executing JavaScript inside a sandboxed iframe.

    Simon Willison··Agents & tooling·newsletter·simonwillison.net
  9. Accelerating Transformer Training with NVIDIA Transformer Engine, Fused Kernels, BF16, FP8, and GPU Benchmarking

    NVIDIA Transformer Engine accelerates workloads by combining fused kernels with FP8 execution.

    Marktechpost··Infrastructure & chips·newsletter·marktechpost.com
  10. GLM 5.2 — B300 vs H200 — Performance per Dollar

    GLM 5.2 benchmarks compare performance per dollar between B300 and H200 hardware.

    SemiAnalysis··Infrastructure & chips·newsletter
  11. Judge denies xAI’s request to block Minnesota ban on ‘nudify’ apps

    A judge denied xAI's request to block a Minnesota ban on 'nudify' apps.

    TechCrunch AI··Business & policy·press·techcrunch.com
  12. Find out what’s new in the Gemini app in July's Gemini Drop.

    Gemini 3.6 Flash and 3.5 Flash-Lite models are now available in the latest update.

    Google (Gemini)··Products & apps·lab·blog.google