AI NewsOctober 1, 2026

Instead of opening a separate Gem to reuse an instruction set, users can now type "/" plus a skill name directly in any Gemini chat.
Google launched Skills in the Gemini app on September 30, free for all users globally. Skills are reusable instructions, for tasks like prepping a presentation or matching a writing style, that can be stacked together in one conversation and triggered from the main prompt box.
Skills replace the older Gems feature, which Google will phase out starting November 17 for personal accounts, migrating existing Gems automatically. Business and education Workspace accounts keep Gems until March and June 2027, respectively.
Key Capabilities:

• Quick access: Typed with a "/" directly in chat, no separate Gem manager needed.
• Stackable: Multiple skills can be combined in a single conversation.
• Automatic migration: Existing Gems convert to Skills with no action needed from users.

Ant Group's Ling-3.1-flash claims scores near GPT-5.6 Sol and Claude Opus 5 on its own charts, but there are no independent tests yet, and the weights aren't public.
Ant Group announced Ling-3.1-flash on September 30. Mixture-of-experts means the model has many specialized sub-networks but only "wakes up" a few per question, which is why 560 billion total parameters only need about 25 billion active at once, keeping it cheaper to run than its full size suggests.
Every number so far, including a claimed 1,673 Elo score and a compiler reportedly built in 17 hours, comes only from Ant's own launch materials. No outside benchmark site has tested it, and open-source weights are promised but not yet released.
Key Capabilities:

• Reported size: 560B total parameters, about 25B active per token, with a context window Ant says can reach 1M tokens.
• Limited trial: A free two-week trial currently caps the context window at 256,000 tokens.
• Unverified scores: All comparisons to rival models come from Ant's own benchmark chart.

Most AI image editors blur or shift pixels after repeated edits. Ideogram 4.5 claims to preserve untouched areas almost perfectly, even after 10 or more rounds.
Ideogram released version 4.5 on September 30, live in its app, API, and partners like Krea, Runway, and ComfyUI. It targets a specific failure mode: each edit normally adds artifacts, pixel shifts, or color drift, and 4.5 is built to resist that across many editing rounds.
It also edits a cropped section of a high-resolution image (up to 24 megapixels) and stitches the result back in without downsizing. Pricing runs $0.03 to $0.22 per image depending on speed, and open weights are planned but not yet released.
Key Capabilities:

• Multi-turn stability: Keeps untouched regions nearly identical across repeated edits.
• High-res editing: Handles source images up to 4,016 by 6,016 pixels.
• Open weights coming: Not available at launch; Ideogram has committed to releasing them later.

Google's first Gemini 4 model claims the top spot on 13 of 19 published benchmarks, but it's going to cybersecurity testers first, not the public.
Google DeepMind announced Gemini 4 Argon on September 30. It targets long, complex work like coding and enterprise research, with an industry-leading 1 million token output limit. Google's own chart shows it beating GPT-6 Astra and Claude Opus 5.5 on most listed tests, including an 88.8% on LABBench 2.
Argon rolls out first to vetted cyber defenders through the Fairwind Program, without cyber guardrails, while a public version with safety restrictions comes later for paid API and AI Ultra users. Google says internal teams already used it to free 300 TiB of data center memory.
Key Capabilities:

• Long output: A 1 million token output limit, four times more than most rival flagships.
• Cyber strength: Ties for first place on vulnerability remediation, scoring 68% on CWE-bench v1.
• Intro pricing: $2 per million input tokens and $10 per million output, with 95% off cached input.