From the source

Together AI expands fine-tuning service with more models, live metrics, and finer controls

Together AI expanded its fine-tuning service to support additional open-weight models including GLM 5.3, DeepSeek-V4-Flash, Kimi K2.7-Code, Qwen 3.8-27B, and Gemma 4.

The update adds live experiment tracking, expert LoRA adapters for Mixture-of-Experts models, early stopping, arbitrary batch size support, and price reductions of 30% to 70% on training for most supported models.

From the source
Today, we're expanding Together Fine-Tuning across that workflow. This release adds support for the latest open-weight models, live experiment tracking, finer controls over training and data processing, and lower prices on selected models.
together.ai

Other stories

Google Research · 2 OctCapability changeToward provably private learning from federated data

Amazon · 1 OctCapability changeServe live, governed data in AI-built apps with Amazon Quick

GitHub · 1 OctCapability changeGitHub Copilot can now interact with desktop apps with computer use

GitHub · 1 OctCapability change

Post on X ↗

ModelsDeepSeek-V4-Flash DeepSeek V4 Flash 0731 Gemma 4 26B A4B Gemma 4 31B GLM-5.1 GLM-5.2 GLM-5.3 Kimi K2.6 Kimi K2.7 CodeGLM 5.3DeepSeek-V4-Flash-0731Kimi K2.7-CodeQwen 3.8-27BQwen 3.6-35B-A3BQwen 3.6-27BQwen 3.5-27BQwen 3.5-9BQwen 3.5-4BQwen 3.5-2BQwen 3.5-0.8BGemma 4-31BGemma 4-26B-A4B

Announced: 11 Sep 2026
Detected: 11 Sep 2026 16:21 UTC
Pricing: dropping the prices of training for most of the models we support, with savings starting from 30% and reaching 70% for models such as the gpt-oss series

Visit source ↗together.ai

GitHub Copilot in VS Code, September 2026 releases

GitHub · 1 OctCapability changeDynamic workflows in Copilot CLI and the Copilot app

Amazon · 1 OctCapability changeSimplify dashboard drill-down with the Amazon Quick Sight hierarchy filter