Command Code Launches Cloud Sandboxes for Running Open Models
Command Code announced Command Code Cloud, a sandbox mode that runs open models in cloud VMs so agents keep working against a GitHub repo on their own branch an…
Global AI Digest
-
-
Command Code Launches Cloud Sessions With Sandboxes
Command Code introduces Cloud sessions with sandboxes, letting users run the agent in the cloud on GitHub repos, keep sessions running when closing their laptop…
-
ElevenLabs Now Available on OpenRouter with 11 Speech Models and 50% Launch Discount
ElevenLabs is now available on OpenRouter with 9 text-to-speech models supporting 90+ languages and 2 speech-to-text models with speaker labels and word-level t…
-
OpenRouter Launches Decision Model Rankings with Spend and Token Share Data
OpenRouter announced a new Decision Model Rankings feature showing spend share and token share across decision model task types like Relevance and Correctness,…
-
Mistral Large 4 Tops AutomationBench and Leads Legal Agent Benchmark Among Open Models
Mistral announced Mistral Large 4, a general-purpose agent model that finished ahead of Kimi K3, MiMo-V2.6-Pro and DeepSeek V4 Pro on AutomationBench's 657 busi…
-
Hugging Face Pollen Robot Gets Custom SBC with Rockchip CPU, Dual NFC and Improved Cooling
Hugging Face's Pollen robot prototype replaced its Radxa-based board sandwich with a fully custom single-board computer built by Seeed, featuring a Rockchip CPU…
-
Requesty Adds Melious as European Provider for Open-Weight Models
Requesty now routes to Melious, a European infrastructure provider serving open-weight models including DeepSeek and GLM variants, all accessible with the same…
-
GMI Cloud Extends Free Qwen 3.8 Access by 7 Days and Raises Rate Limits
GMI Cloud extends free access to Qwen 3.8 Max and Flash for 7 more days, increases rate limits across all three models, and launches a community build contest w…
-
OpenCode Team Ships iOS App Through Slack Prompting Without Running It Locally
OpenCode's team describes an unusual development workflow for their iOS app: no one runs the repo locally, all coordination happens by prompting in Slack and re…
-
LiteLLM Adds Day 0 Support for Mistral Large 4
LiteLLM announces day 0 support for Mistral Large 4 (Le Chonk) - a 1T parameter natively multimodal model with 49B active parameters, claiming best open-weights…
