Keras Kinetic has a new alpha release: v0.0.2! Including a new docs website: http://
kinetic.readthedocs.io Kinetic is my favorite new release from the Keras team: a super simple Modal-like API to run training jobs on TPU.
RESEARCH
-
Keras Kinetic v0.0.2 Alpha: Modal-like TPU Training API
By
–
-
Missing Verifiers: Key Primitive for Autonomous AI Agents
By
–
feels like we have all the primitives for building powerful autonomous agent model
task management
tool use/auth
memory & context management
self improvement
UI/UX what’s missing? verifiers (for self improvement) we have this in coding, which is why coding agents work so well -
METR Evals: Best Public Signal for AI Capability Progress
By
–
METR's evals are honestly the best public signal we have for actual capability gain right now, glad they're getting more airtime.
-
Agent-First Benchmarks: Evaluating Variable Step Protocols
By
–
Agent-first benchmarks are something! Curious what the protocol looks like for agents that take wildly different numbers of steps.
-
AI Often Judged on Outdated Models, Skewing Research Results
By
–
This points to a broader issue where AI is often being judged based on extremely outdated models. Any time someone sends me a study showing AI failing at something, the first thing I do is look at the models they used… more often than not, they’re extremely outdated. We need
-
AI Becomes Core Growth Engine for Telecom Operations
By
–
AI has become the core growth engine for telecom operations, networks, and services.
— NVIDIA (@nvidia) 27 avril 2026
NVIDIA’s 2026 State of AI in Telecommunications Report shares data from over 1,000 global professionals on how companies are investing in, deploying, and benefiting from AI.
Learn about the… pic.twitter.com/dSpTchCgBhAI has become the core growth engine for telecom operations, networks, and services. NVIDIA’s 2026 State of AI in Telecommunications Report shares data from over 1,000 global professionals on how companies are investing in, deploying, and benefiting from AI. Learn about the
-
Xiaomi MiMo-V2.5 Pro: New Open Source SOTA LLM
By
–
New Opensource SoTA contender enters the arena Xiaomi MiMo-V2.5 Pro
– 1.02T Total Params / 42B Active Params
– Base and Instruct versions Xiaomi MiMo-V2.5
– 310B Total Params / 15B Active Params
– Base and Instruct versions MIT License Opensource AI just keeps getting better -
Bullshit Benchmark: Evaluating AI Model Reliability
By
–
GitHub: https://
github.com/petergpt/bulls
hit-benchmark
… DataViewer: https://
petergpt.github.io/bullshit-bench
mark/viewer/index.v2.html
… -
DeepSeek v4 Underperforms on BullshitBench Reasoning Tasks
By
–
BullshitBench: sorry to say but DeepSeek v4 did really badly, towards the bottom of the table, whether it is high or low reasoning.
-
AI Models Learn Self-Improvement Without External Rewards
By
–
Can an AI teach itself to reason better without any outside reward? Researchers from CUHK, Shenzhen, SJTU, and CUHK present SePT. They let a language model generate its own reasoning examples by using "low-temperature" (more focused) responses, then train on that new data in a