tensorfolk

Google unveils Gemini 4 Carbon, now #1 on Artificial Analysis Index v5, with access limited to cyber defenders

Big news

Google DeepMind announced Gemini 4 Carbon, a frontier model it pitches on cybersecurity and long-horizon agent work. It scores 59 on Artificial Analysis Intelligence Index v5, one point above Claude Opus 5.5. For now only selected defenders in Google's Fairwind Program can use it, with broader API access promised later at an introductory $2/$10 per million tokens.

Why it matters

A new top frontier model gated by cyber risk. Teams planning on Gemini should watch for API access and note that its hallucination rate is higher than Argon's.

Source: Read the full story on Google

geminigoogle-deepmindfrontier-modelscybersecuritybenchmarks

More in Foundations & Models

Foundations & Models / Model Releases & Landscape

OpenAI ships GPT-6.1 Terra, a cheaper mid-tier model built for computer use

OpenAI released GPT-6.1 Terra on October 26 to replace GPT-5.6 Terra, pitching it at everyday screen-driven work like form filling and hopping between apps. API pricing drops to $1 input and $5 output per million tokens, with a 1.05M-token context and up to 128K output. It is available in ChatGPT paid plans, Codex and the API as gpt-6.1-terra.

Why it matters: Computer-use agents get much cheaper to run; Artificial Analysis puts cost per task at under a third of the previous Terra.

How to try it

GPT-6.1 Terra is OpenAI's lower-cost model for visual reasoning, computer use and writing, with a 1.05M-token context window and 128K max output.

  1. In ChatGPT on a paid plan, or in Codex, choose GPT-6.1 Terra from the model picker.
  2. In the API, set the model name to gpt-6.1-terra in your existing Responses or Chat Completions calls.
  3. Keep prompts under 272K input tokens where you can: longer requests are billed at a higher rate.
  4. Use Batch or Flex for non-urgent jobs, which cost half the standard price.
Open the docs
Big newsRead on OpenAI
Foundations & Models / Model Releases & Landscape

Moonshot launches Kimi K3.1 Pro, with open weights promised by Nov 4

Moonshot shipped Kimi K3.1 Pro, a further-trained Kimi K3, in its apps and API, plus a Highspeed variant said to be 2-3x faster. Moonshot reports 65.7% on Terminal-Bench 4.0, near the top closed models, and a big drop in claim-level hallucination. It says full weights will be published by November 4.

Why it matters: A near-frontier coding and agent model that may soon be self-hostable; scores are vendor-reported, so test on your own tasks.

Big newsRead on The Next Web
Foundations & Models / Small & On-device Models

Liquid AI releases LFM3: six open-weight ~2B models for text, vision, audio, omni and decisions

Liquid AI's LFM3 family has six roughly 2B-parameter open-weight models sharing one backbone trained on 22T tokens, which mixes short convolutions with sliding-window and global attention. Variants cover 256K-context text, image and video input, audio, omni input, and two decision models. Weights use Liquid's LFM Open License v1.0.

Why it matters: A capable on-device family where the 4-bit text model fits in about 1GB, with day-one support in llama.cpp, MLX, vLLM, Ollama and LM Studio.

How to try it

LFM3-1.7B is the text model of the family, small enough to run on a laptop. Liquid publishes GGUF builds for llama.cpp and recommends low-temperature sampling for chat, tool use and coding.

  1. Install llama.cpp (or use Ollama, LM Studio or MLX, which Liquid lists as supported).
  2. Run the official GGUF build straight from Hugging Face with the command below.
  3. For Python, load LiquidAI/LFM3-1.7B in a Transformers text-generation pipeline with trust_remote_code=True.
bash
llama-cli -hf LiquidAI/LFM3-1.7B-GGUF \
--temp 0.1 --top-k 50 --repeat-penalty 1.1
Open the docs
Big newsRead on Liquid AI