Skip to content

AI News

Each item is checked against its primary source and gets a verdict. Pick a topic to filter.

November 2026

Nov 9
Google adds TPU 8tx and 8i Max to its eighth-generation line
Vendor-reported · Model releases · Adoption and cost
Nov 9
Anthropic releases Claude Fable 5.2 and Mythos 5.2
Vendor-reported · Model releases · Agents and coding · AI security
Nov 9
Meta Muse Ember: cheap agent model, claims not yet checkable
Pending · Model releases · Adoption and cost
Nov 8
llama-modchip claims DeepSeek V4 Flash at 9 to 14 tokens/s on a 64 GB CPU-only machine
Vendor-reported · Model releases · Research
Nov 7
xAI turns on invisible text watermarking in the Grok API by default
Confirmed · Model releases · AI security
Nov 5
OpenAI GPT-6 Cyber reports 48.5% on ExploitGym
Vendor-reported · Model releases · AI security
Nov 5
Thinking Machines launches Inkling 2 Small with closed weights, Apache 2.0 promised
Confirmed · Model releases · Benchmarks
Nov 5
DeepSeek says Harness has 1.54M daily active users
Vendor-reported · Agents and coding · Adoption and cost
Nov 4
OpenAI to acquire Pierre, maker of Git storage and diff libraries used in Codex
Confirmed · Agents and coding · AI in software delivery
Nov 4
OpenAI reports its agent pipeline built working exploits for 948 Linux kernel privilege-escalation bugs
Vendor-reported · Agents and coding · AI security
Nov 4
Cognition SWE-2 Flash: cheap, fast coding model with vendor-run benchmarks
Vendor-reported · Model releases · Agents and coding
Nov 3
Collinear and Hugging Face publish CWE-bench v2, 100 defensive patch tasks
Vendor-reported · Benchmarks · AI security
Nov 2
Goldman Sachs reportedly limits Claude Code and Copilot to engineers
Pending · Agents and coding · Adoption and cost
Nov 2
MIT CSAIL: LLM edits a small model's weights and lifts MATH-500 from 7.9% to 48.1%
Vendor-reported · Research

October 2026

Oct 29
Mistral Large 4 open weights: ranking claims differ from Artificial Analysis
Vendor-reported · Model releases · Benchmarks
Oct 28
Google releases Gemini 4 Carbon to cyber defenders only
Vendor-reported · Model releases · AI security
Oct 26
GPT-6.1 Terra: $1/$5 per million tokens, Artificial Analysis confirms the cost drop
Confirmed · Model releases · Benchmarks · Adoption and cost
Oct 26
Co-Design Bench finds top coding model completes 26.4% of tasks
Confirmed · Agents and coding · Benchmarks
Oct 23
Moonshot publishes Kimi K3.1 Pro weights; its Artificial Analysis rank fell under index v5
Vendor-reported · Model releases · Benchmarks
Oct 12
FrontierCode leaderboard: best agent scores 52%, leaders fail over 40% of tasks
Confirmed · Agents and coding · Benchmarks