Molmo2-4B is a 4.85B video model from December 2025 with 19.42 GB of weights. LFM2.5-VL-3B does images only in under 3.3 GB.

Two models called Omni, pointed in opposite directions: one writes video at $0.07/sec, one reads an hour of it for $0.20.

Ming 3.0 is five MIT audio checkpoints; Gemini Omni 1.2 Flash bakes a voice into 15 seconds of video. Where they overlap, and ...

Gemini Omni 1.2 Flash charges about $0.07 a second. InternLumina-U2 is 67.6 GB of Apache-2.0 weights. What each can do, and ...

A 15-second Gemini Omni 1.2 Flash clip costs about $1.05; a 1K still from Nano Banana 2.1 costs $0.0336. Which to buy first ...

A 5 November post says Claude Sonnet 5.5 at Extra effort is serving the next Claude model. Both its tests pass on the model Anthropic already ships.

A prompt injection in a GitHub issue title let an outsider poison Cline's CI cache and steal release tokens. What is documented, and what is not.

In February 2026, Naver, Kakao and Karrot restricted OpenClaw on work devices. Only Kakao confirmed it and no leak at any of ...

GPT-6 Luna publishes price, latency, context and an index. ERNIE-5.5-Flash publishes none of the four — a test of what the ...

Both open repos are downloadable, but only GLM-5.3 Flash is finished, multimodal and MIT-licensed. What ERNIE-5.5-Flash-Base ...

ERNIE-5.5-Flash-Base is raw weights you post-train; Qwen 3.8 Flash is a finished API model. The benchmarks that don't compare ...

Base, a 280B pretraining checkpoint, on 5 November 2026 — but the licence excludes the EU, UK and South Korea.