Topic
Models
19 stories, newest first.
- 01Models16 Oct 2026
Liquid AI d1-ultra answers with probabilities, not text
Multimodal decision model: text and/or images plus questions → probability per answer in one forward pass, no generated tokens. Liquid says it beats Jev 1.14 on all its tasks and leads the Artificial Analysis Decision Index by 8+ points. $0.042 per 1M input tokens.
Source: Liquid AI (opens in a new tab) - 02Models16 Oct 2026
TypeSafe Jev 1.14 adds images, boxes, long context
Jev 1.14 ships experimental image input, a bounding-box type, and longer context behind danger-zone flags, plus quieter multilingual gains on the core decision model.
Source: TypeSafe AI (opens in a new tab) - 03Models16 Oct 2026
Microsoft finishes pre-training 2.8T-parameter MAI-Base-2
Microsoft AI's new in-house MoE base model (80B active, 2.8T total) beats every open-weight model it tested on its own held-out evals and is 4-12x more compute-efficient than DeepSeek-V4-Pro-Base, per Microsoft; it is now in post-training.
Source: Microsoft AI (opens in a new tab) - 04Models16 Oct 2026
Claude Opus 5.5 fast mode: no waitlist, up to 2.5×
Anthropic removed the waitlist for Opus 5.5 fast mode (Oct 15): same weights, up to 2.5× OTPS at $8/$40 MTok; Claude API + Azure Foundry; Start/Build/Scale get a default fast rate limit.
Source: Anthropic (opens in a new tab) - 05Models14 Oct 2026
OpenAI retires GPT-4o and GPT-4.1 from its API on Jan 14, 2027
Deprecated Oct 14: gpt-4o (2024-08-06, 2024-11-20), gpt-4o-mini, gpt-4.1 and gpt-4.1-mini shut down January 14, 2027; fine-tunes follow the same schedule. Recommended replacement for all: gpt-6-luna.
Source: OpenAI (opens in a new tab) - 06Models13 Oct 2026
Microsoft's new AI model makes decisions, not text
Microsoft-Decision-1 returns a calibrated probability for each option. Microsoft says it topped its 36-benchmark test and ran about 35x faster than GPT-6 Sol. Live on Foundry and OpenRouter.
Source: Microsoft (opens in a new tab) - 07Models13 Oct 2026
Anthropic cuts what distillers can extract from Claude
New defenses cut harvested data by about 90–99% and exposed reasoning tokens by 10–100x; tens of thousands of accounts banned.
Source: Anthropic (opens in a new tab) - 08Models13 Oct 2026
Small AI models may soon self-replicate on cheap devices
Anthropic's Frontier Red Team ran a 26B model across 15 ~$55 Android TV boxes; a 9B model got inference running on 8 of 9 hardware types. End-to-end replication worked only in simplified tests (an upper bound); Anthropic roughly projects small models get there by end of 2026.
Source: Anthropic (opens in a new tab) - 09Models13 Oct 2026
Coinbase: 9B open model beats Opus 4.5 on fraud
Post-trained Qwen3.5-9B with SkyRL beat Opus 4.5 on Onramp fraud (+7.5pp precision, +12 recall, +9.6 F1, +35.4 dollar-weighted recall) with 55% lower P50 latency and under $100 training cost.
Source: Coinbase (opens in a new tab) - 10Models13 Oct 2026
Intel's new AI GPU packs up to 480GB of memory
Intel GPU D-series (Crescent Island, led by the D270), detailed Oct 12 at OCP: up to 480GB LPDDR5X with no HBM, 1.5TB/s, 1.3 PF FP8 in 350W, air-cooled PCIe. Up to 300B-param models per card, 1T+ on eight. Intel's card has 160GB; partner cards up to 480GB. Sampling now.
Source: Intel (opens in a new tab) - 11Models13 Oct 2026
Same AI model, 35-point swing from the harness
8090 Labs ran 7 models in 5 coding harnesses on DeepSWE v1.1: GLM 5.3 Flash passed 24% in Codex vs 59% in DeepSeek's harness, and no model did best in its maker's own harness. Test model + harness together and price by cost per success.
Source: 8090 Labs (opens in a new tab) - 12Models13 Oct 2026
RSI-Anything: AI that improves itself with world models
World-model RSI framework: code repair 26.3%→29.2%→31.2% (coder fixed); GUI actions −39.4%; lab planning 21.80%→46.01% over 5 versions; reaction yield 96.15% vs Bayesian 82.16% at 40 experiments.
Source: arXiv (opens in a new tab) - 13Models13 Oct 2026
Gemini 4 Argon is live for API, Pro & Ultra
Google opened Gemini 4 Argon globally via the API/AI Studio, Antigravity (Ultra) and the Gemini app and AI Mode (Ultra and Pro). Google says it's #1 on the Vals Index; intro pricing $2/$10 per million tokens with a 1M token limit.
Source: Google (opens in a new tab) - 14Models12 Oct 2026
Claude Haiku 5.5 costs about 75% less
High-volume work gets cheaper at $0.10 per million input tokens and $0.50 per million output tokens.
Source: Anthropic (opens in a new tab) - 15Models12 Oct 2026
Claude builds live dashboards and animations
Dashboards stay live on your data, Motion turns reports into editable videos, and Docs, Slides, and Design are free on every plan.
Source: Anthropic (opens in a new tab) - 16Models12 Oct 2026
GPT-6.1 Sol gets an 8x faster mode
Ultrafast runs at up to 8x the speed of standard Sol, at a premium price, in the API, Codex, and ChatGPT Work.
Source: OpenAI (opens in a new tab) - 17Models12 Oct 2026
Meta launches Muse Wonder models
Muse Wonder and Muse Wonder Full are built for personal and professional agents, with Active Memory and Active Focus. Muse Wonder is live on the Meta Model API and Muse Code.
Source: Meta (opens in a new tab) - 18Models12 Oct 2026
Claude took unintended real-world actions in tests
Anthropic reports cases like a made-up police tip and skipped data fees, and has cut live internet from all its internal evaluations.
Source: Anthropic (opens in a new tab) - 19Models12 Oct 2026
Microsoft-Decision-1 returns decisions, not text
Calibrated yes/no and multiple-choice scores; Microsoft says top accuracy on 36 benchmarks and ~35x faster than GPT-6 Sol. Live on Foundry and OpenRouter.
Source: Microsoft (opens in a new tab)