Blog

K-EXAONE 2.5 infographic: LG free 750B model vs DeepSeek V4.1 Flash, 512K context, 59.21 SWE-Bench Pro

AI Tools & Updates

K-EXAONE 2.5: LG's Free 750B Model vs DeepSeek

LG opened K-EXAONE 2.5 as a free API on Nov 9, 2026: 750B parameters, 31B active per token, a 512K context window, and agentic benchmark scores roughly level with DeepSeek V4.1 Flash and GLM 5.3. Here is what it does, what it costs, and the catch.

Inkling 2 Small infographic: Thinking Machines 594B model, 17B active per token, 31 on the AA Intelligence Index

AI Tools & Updates

Inkling 2 Small Explained: Thinking Machines' 594B Model

Thinking Machines launched Inkling 2 Small on Nov 5, 2026: 594B parameters, 17B active per token, 31 on the Artificial Analysis Intelligence Index, at $0.30/$1.00 per 1M tokens. The weights are still closed, with an Apache 2.0 release promised next week.

GPT-6 Cyber infographic: OpenAI security model, 48.5% ExploitGym vs 32.1% for GPT-5.6 Cyber, 948 Linux exploits, $20 / $100 per 1M tokens

AI Tools & Updates

GPT-6 Cyber Explained: OpenAI's Security Model

OpenAI launched GPT-6 Cyber on Nov 5, 2026: 48.5% on ExploitGym, 948 Linux exploits built, $20/$100 per 1M tokens. Benchmarks, price and who can actually use it.

Gemini 4 Carbon infographic: 88.2% SEC-bench Pro, 54.8% ExploitGym, 71.0% CWE-bench

AI Tools & Updates

Gemini 4 Carbon Explained: Google's Cybersecurity Model

Google launched Gemini 4 Carbon on Oct 28, 2026: a cybersecurity-focused frontier model with 88.2% on SEC-bench Pro, 71.0% on CWE-bench, $2/$10 pricing, and first access only for vetted cyber defenders via the Fairwind Program.