AI Daily · Sep 24, 2026: Claude Marketplace Launches, Muse Unveils Multiple Updates
Claude Marketplace launches today with tools, agents, and partner connectors; Meta unveils Muse Realtime Avatar and rolls out new models with free trials, covering voice, translation, and more, worth attention for developers and users.
Pick a date
-
Claude Marketplace Launches with Tools, Agents, and Partners
Claude announced the launch of Claude Marketplace, where users can discover tools, agents, and expert partners, including connectors like Slack and Notion, and products from companies like Cursor and CrowdStrike.
-
Meta unveils Muse Realtime Avatar, real-time embodiment tech
Meta unveiled Muse Realtime Avatar, a real-time embodiment technology that turns Muse Realtime Voice into expressive avatars, enabling real-time conversations in Muse; it came out ahead in live-call comparisons against two leading commercial avatar systems.
-
Tencent Launches Hy-MT2 Translation App with Offline Support
Tencent Hy Translation, powered by Hy-MT2, supports 33 languages, 5 Chinese minority languages/dialects, voice and photo translation, and full offline use, now live in 12 countries.
-
Tencent Hy Image 3.5 Preview Now on GMI Cloud at $0.024/Image
Hy Image 3.5 preview is live on GMI Cloud at $0.024 per image, claimed 8.8x cheaper than GPT Image 2 and 5.6x cheaper than Nano Banana Pro.
-
Xiaomi MiMo-V2.6 Series Gets Limited-Time Discounts Starting Sept 25
On B.AI, Xiaomi MiMo-V2.6-Flash gets a 90% discount and MiMo-V2.6-Pro gets a 50% discount starting Sept 25, 15:00 SGT.
-
Xiaomi MiMo-V2.6 Pro Claims Top Open-Source Spot with AI Index 46
MiMo-V2.6 Pro matches Grok 4.7 with an Artificial Analysis intelligence index of 46, topping open-source and domestic models; Flash excels in AutomationBench and Terminal Bench for high-throughput agent tasks.
-
OpenCode Server Fixes Code Execution Vulnerability in Versions 1.14.30-1.18.21
OpenCode Server has a code execution vulnerability affecting versions 1.14.30 through 1.18.21; users should update to the latest version, reported by christophetd and Datadog.
-
Command Code Plan Handles 1.5T Tokens; Claude Pro Fails at 5 Messages
A user reports Command Code's goat plan handled 1.5 trillion tokens without issue, while Claude Pro + Opus 5.5 lasted only 5 messages before hitting limits.
-
Requesty Data Shows Flash Models Onboard 3x More Teams in First Week
According to Requesty, adoption rates 7 days after launch: DeepSeek V4.1 Flash 11%, Gemini 3.8 Flash 8%, Claude Opus 5 8%, GPT-6 Astra 4%, Claude Fable 5.1 2%; flash tier models onboard about 3x more teams in week one.