Models
114 articles RSS
OpenAI's GPT-6.2 Astra and Sol Beat Benchmarks and Cross a Self-Improvement Threshold, but Power Users Call Astra Unusable
OpenAI's GPT-6.2 Astra tops independent benchmarks and crosses a new AI self-improvement threshold, but early power users call it unusable for long agentic work.
A 2024 Hallucination Bound Now Anchors Three 2026 Papers on LLM Training Data, One Extending It to Scientific Discovery
A 2024 theorem tying LLM hallucination to once-seen training facts has drawn three 2026 papers, one linking it to scientific discovery and another challenging standard deduplication practice.
Arena, Artificial Analysis and Terminal-Bench Disagree on Which AI Model Leads This Month
Arena, Artificial Analysis and Terminal-Bench each credit a different model at the top, and researchers say the disagreement traces to what each benchmark actually measures.
Perplexity Releases Decider v1.3 and Flash v1.3 Weights Under Apache 2.0; Artificial Analysis Decision Index Lists v1.3 Above Jev 1.14, Flash Below
MarkTechPost reports Perplexity's open-weight Decider v1.3 and Flash v1.3 ship under Apache 2.0 on Qwen bases; Artificial Analysis's Decision Index scores v1.3 at 70.41, Jev 1.14 at 66.23 and Flash v1.3 at 64.07.
Artificial Analysis Scores Google's Gemini 4 Hydrogen 51 Against Argon's 50 at High Thinking, With $1.87 per Index Task Versus $4.97
Google's Gemini 4 Hydrogen costs $1.50 and $7.50 per million tokens. Artificial Analysis scores it 51 at high thinking against Argon's 50; Google Cloud lists 3.8 Flash at half that price through December 31.
How Much Notice Developers Get Before AI Models Retire: OpenAI, Anthropic and Google Cloud Publish Minimums, GitHub Copilot Dates Imply About a Month
OpenAI, Anthropic and Google Cloud publish different minimum notice periods for retiring AI models; the Herald's date arithmetic puts GitHub Copilot's notices at 31 to 32 days, and one OpenAI notice ran about three months.
Google Launches Gemini Decisions API With Up to 1,000 Questions per Request at $0.10 per Million Input Tokens; Its Benchmark Table Is Self-Reported
Google says its Gemini Decisions API takes up to 1,000 questions per request at $0.10 per 1M input tokens; Artificial Analysis' Decision Index data scores it between Jev 1.14 and OpenAI's entry.
GPT-Image-2.6 Sunburst and Flare reach OpenAI's API; the API reference says GIF needs preview access and 2.6 model pages list Batch as not supported
OpenAI's November 4 API release adds a complexity setting, SVG and GIF output and markup edits; the API reference says GIF requires preview access, the 2.6 pages list Batch as not supported, and gpt-image-2 is deprecated for removal on February 10, 2027.
Meta's Muse Ember Scores 36 at $0.14 per Task at Max Reasoning on Artificial Analysis, Which Reports 68.1% on AutomationBench-AA Versus Meta's 37.8%
Meta lists Muse Ember at $0.10 input and $0.50 output per million tokens for prompts up to 200K tokens; Artificial Analysis scores it 36 at $0.14 per task (max) and gives AutomationBench-AA 68.1% where Meta's table shows 37.8%.
Gemini Omni 1.2 Flash Is Generally Available With 15-Second Clips, and Google Cloud Lists Video Output at $12 per Million Tokens, Down From $17.50
Google says Gemini Omni 1.2 Flash is generally available with clips up to 15 seconds and 720p video at $0.07 per second; three Artificial Analysis boards read November 9 list 1.1, not 1.2.
Nous Research Releases Arete-27B, a Qwen3.8-27B Fine-Tune That Reports a 0.9% Propaganda Rate on Its Own 803-Prompt Eval, Graded by Grok 4.7
Nous Research released Arete-27B, a Qwen3.8-27B fine-tune. Its card reports propaganda in 0.9% of answers versus 22% for the base model on Nous's own 803-prompt eval, graded by Grok 4.7; Nous's pages give differing dataset sizes.
What Four U.S. Coding-Model Releases Disclose: Licenses and Weight Access Differ, and Each Release's Benchmark Table Is the Vendor's Own
Beam and Ripple-65B list Apache 2.0, Muse Spark Open adds a $20 million Model-as-a-Service condition, and SWE-2 Flash keeps its weights private; each benchmark table is the vendor's own, run with differing setups.