Is Day 17 gone? Did Tibo fall asleep and forget?
Model intelligence feed
Model intelligence.
Follow the changes. Understand what comes next.
Editor’s pickOther providers
Valve's Answer to AI Slop and Clone Games on Steam: We Won't Police Developers' Tools
At Steam Indie Connect in Korea, Valve staff addressed AI slop and clone games head-on: no gatekeeping on AI tools, and the best defense against clones is community and steady updates.
Latest stories341
Newest published181182183184185186187188189190191192
Ai2 OlmoEarth 1.3 Lets You Spot Land Change Without Training a Model
On October 22, 2026 (Beijing time), Ai2 released OlmoEarth v1.3: satellite embeddings, global annual 10-meter features, and a browser Embeddings Explorer. This guide separates three jobs—year-flip comparison, a small classifier in the Explorer, and full fine-tuning in Studio—and notes coverage and license limits. We have not tested the Explorer yet.
OlmoEarth v1.3 / Embeddings Explorer / OlmoEarth StudioRead story
Baseten Mammoth-1 and Elephant-1 Are Free on Vercel Through October 27
Baseten’s Base Labs released mammoth-1 (27B) and elephant-1 (9B) decision models—Choice, Score, and Boolean questions that return probability distributions, not generative coding LLMs. They are free on Vercel AI Gateway through October 27, 2026 (also free for a week on Baseten Model APIs), open-source on Hugging Face as baselabs/mammoth-1 and baselabs/elephant-1, and post-trained from Qwen3.5-27B / Qwen3.5-9B. Rankings are Baseten’s claims citing Jev and Artificial Analysis.
Mammoth-1 / Elephant-1 / System One / Vercel AI GatewayRead story
Canva's First Image Model Craft Hits Dream Lab and a Public API
Canva Research’s first image model, Craft, is available to Canva users globally through Dream Lab and to developers via a public image-generation API with model craft_v1_2k. DesignArena lab and model rankings are Canva’s own citations. Canva MCP for Craft is listed as Coming Soon.
Craft / Dream Lab / craft_v1_2kRead story
OpenAI Shuts Down GPT-4 Turbo, o1 and Other Legacy API Models on October 23
On October 23, 2026 (UTC day), OpenAI shut down a batch of legacy API model snapshots first announced on April 22, 2026—including gpt-4-turbo, gpt-4-0613, gpt-4o-2024-05-13, o1, o3-mini, o4-mini, gpt-image-1, and listed fine-tunes. Official substitutes point mainly to gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna, or gpt-image-2.5-*. We have not called these endpoints to verify the shutoff. This article covers only that October 23 batch.
GPT-4 Turbo / o1 / legacy API snapshots (Oct 23 shutdown)Read story
Codex CLI 0.164.0 Adds Ten /presets Slots and Inline /model Switches
OpenAI shipped Codex CLI 0.164.0 (GitHub rust-v0.164.0; npm latest). Developer highlights include ten /presets slots, inline /model and /permissions arguments, Kitty/Sixel image preview in the fullscreen transcript, enterprise MCP auth options, remote credential masking, and a default-on Amazon Bedrock setup wizard. We have not tested this build. It is a CLI release—not a usage-quota or Day 18 reset update.
Codex CLI 0.164.0Read story
Gemini API CLI Preview Needs an API Key — It Isn't Gemini CLI
Google AI Studio published an experimental preview of Gemini API CLI (binary name gemini-api): a stateless CLI so shell-capable coding agents can call Gemini models, media generation, file analysis, and managed agents with an API key. We have not measured it. Do not confuse it with the separate Gemini CLI covered in our usage-limits guide.
Gemini API CLI (gemini-api)Read story
Framer MCP Connects Claude, Codex, and Cursor — Publish Still Needs a Go-Ahead
On about October 22–23, 2026 Beijing time, Framer shipped Framer MCP: a remote Model Context Protocol server so agents can work on the canvas, CMS, and project settings without keeping the Framer editor open. Live-site updates still require an explicit publish. We have not tested the flow.
Framer MCPRead story
Which IBM Granite Speech Model Fits Your Recording?
On about October 21, 2026 Beijing time, IBM posted Granite Speech 5.0: a new 2B speech-language model plus updates to Turbo CTC and 4.1-2b-plus. This guide maps which recording type points to which model, keeps English-to-Mandarin as a translation direction only, and notes Apache 2.0. We have not measured accuracy, speed, or hardware needs.
Granite Speech 5.0/granite-speech-5.0-2b/granite-speech-5.0-470m-turboctc/granite-speech-4.1-2b-plusRead story
FastH3 Omni on a Local GPU — Thresholds, Speed Caveats, License Limits
FastVideo and Nuva Lab open-sourced FastH3 Omni for reference-guided video with synced audio. This guide maps official local GPU recipes (DGX Spark, RTX 5090, FP8/4090/ComfyUI), keeps the 5.6–11.9× claim tied to eight GB200 matched tests, and spells out the MiniMax-H3 Community License exclusions. We have not measured quality or speed.
FastH3 OmniRead story
TorchNeuron Public Beta Puts PyTorch on Trainium — What Runs As-Is
AWS opened TorchNeuron as a Public Beta native PyTorch backend for Trainium2 and Trainium3. Device placement can move to neuron with eager mode, torch.compile, and standard APIs; compile backends, unsupported ops, and NKI custom kernels still need attention. Open source is not free cloud compute, and We have not benchmarked cost or speed.
TorchNeuron / Trainium2 / Trainium3 / torch-neuronxRead story
MiniMax M3.1 Flash Open Weights Are on Hugging Face
As of 08:59 Beijing time on October 23, 2026, open weights for MiniMax-M3.1-Flash are on Hugging Face. The model card describes about 428B total MoE parameters with 23B active per token, multimodal inputs, and up to 1M context under the MiniMax Community License. Pricing and API context remain in the main article.
MiniMax M3.1 FlashRead story
Coding Agent Index Moves to v2.0 With FrontierCode — Claude Code + Opus 5.5 Tops the New Board
On October 23, 2026 at 08:17 Beijing time, Artificial Analysis launched Coding Agent Index v2.0: Cognition's FrontierCode v1.1 replaces DeepSWE, SWE-Atlas-QnA becomes SWE-Atlas-QnA-AA, and Terminal-Bench 4.0 stays. Claude Code with Claude Opus 5.5 (max) leads at 61.7 on the new board; Devin Fusion CLI with Opus 5.5 Max + GPT-6 Luna High sits near 61.5. Index totals should not be hard-compared with older versions. We have not re-run the index.
Artificial Analysis Coding Agent Index v2.0Read story
From the Webmaster
Personal experiences, thoughts and notes.