https://ai-news.ypetit.net/sources/together-ai/ Latest publications from Together AI, aggregated by AI News from the official source. en Fri, 23 Oct 2026 19:08:30 GMT https://www.together.ai/blog/together-post-traininghttps://www.together.ai/blog/together-post-trainingMon, 19 Oct 2026 00:00:00 GMTblogPost-train open LLMs like Qwen, GLM and Kimi into custom models you own, with SFT or RL. You write the training loop. Together runs the hard infrastructure. https://www.together.ai/blog/meta-muse-spark-openhttps://www.together.ai/blog/meta-muse-spark-openThu, 15 Oct 2026 00:00:00 GMTblogMeta Muse Spark Open, Meta's open-weight model for agentic coding, is live on Together AI with a 1M-token context window on serverless and dedicated infrastructure. https://www.together.ai/blog/expanding-our-enterprise-inference-capacity-with-ibm-cloud-and-nvidiahttps://www.together.ai/blog/expanding-our-enterprise-inference-capacity-with-ibm-cloud-and-nvidiaTue, 06 Oct 2026 00:00:00 GMTblogEnterprises can now run open models at production scale on a dedicated B300 inference cluster, built by Together AI, IBM Cloud, and NVIDIA https://www.together.ai/blog/together-link-frontier-quality-open-models-in-the-harness-you-already-usehttps://www.together.ai/blog/together-link-frontier-quality-open-models-in-the-harness-you-already-useMon, 05 Oct 2026 00:00:00 GMTblogTogether Link brings frontier open models like GLM 5.3 and Kimi K3 into the coding agent your team already uses, cutting model spend by over 50%. https://www.together.ai/blog/how-to-train-your-own-jevhttps://www.together.ai/blog/how-to-train-your-own-jevWed, 23 Sep 2026 00:00:00 GMTblogWe just launched our own Jev-like classifier, together/Tev1-4B-experimental, on top of Qwen3.5 4B on Together’s serverless platform. In this blog post we’ll show you how to fine-tune your own version! https://www.together.ai/blog/canary-rollouts-upgrade-models-in-production-without-downtimehttps://www.together.ai/blog/canary-rollouts-upgrade-models-in-production-without-downtimeTue, 22 Sep 2026 00:00:00 GMTblogA hard model swap exposes every user at once, and rolling back means cold-starting the old deployment under pressure. Here's how staged traffic ramps, metric gates, and automatic rollback work on dedicated inference. https://www.together.ai/blog/global-fintech-scales-coding-agent-traffic-with-dedicated-model-inferencehttps://www.together.ai/blog/global-fintech-scales-coding-agent-traffic-with-dedicated-model-inferenceFri, 18 Sep 2026 00:00:00 GMTblogInside a global bank's shift to self-serve dedicated inference: how Together's DMI gave engineering teams direct control over scaling, models, and testing. https://www.together.ai/blog/migrating-from-closed-to-open-source-modelshttps://www.together.ai/blog/migrating-from-closed-to-open-source-modelsWed, 16 Sep 2026 00:00:00 GMTblogMoving from closed to open source models can take weeks, not years. A five-stage playbook: discover, evaluate, adapt, decide, and production. https://www.together.ai/blog/together-ai-expands-fine-tuning-service-with-more-models-live-metrics-and-finer-controlshttps://www.together.ai/blog/together-ai-expands-fine-tuning-service-with-more-models-live-metrics-and-finer-controlsFri, 11 Sep 2026 00:00:00 GMTblogTogether Fine-Tuning adds the latest open-weight models, live experiment tracking, Expert LoRA, early stopping, tokenized dataset previews, pre-flight validation, and lower training prices on selected models. https://www.together.ai/blog/introducing-preemptible-compute-the-same-compute-half-the-pricehttps://www.together.ai/blog/introducing-preemptible-compute-the-same-compute-half-the-priceThu, 10 Sep 2026 00:00:00 GMTblogTogether GPU Clusters now supports preemptible compute: the same GPU capacity at a flat 50% of the on-demand rate, with a five-minute drain window. https://www.together.ai/blog/to-infinity-and-beyond-thunderkittens-now-on-nvidia-vera-rubin-nvl72https://www.together.ai/blog/to-infinity-and-beyond-thunderkittens-now-on-nvidia-vera-rubin-nvl72Thu, 10 Sep 2026 00:00:00 GMTblogWe ported ThunderKittens to NVIDIA's Vera Rubin NVL72 and rebuilt our NVFP4 GEMM around the new hardware, taking it from 42% of roofline to over 22 PFLOPS — competitive with cuBLAS and CuTe DSL. Here is what changed in the ISA and how we used it. https://www.together.ai/blog/the-open-source-ai-stackhttps://www.together.ai/blog/the-open-source-ai-stackWed, 09 Sep 2026 00:00:00 GMTblogA deep dive into the open model AI stack — model, inference, gateways and routers, harness, and tools — and how keeping each layer independent lets you swap in a new open model in minutes instead of rebuilding your workflow. https://www.together.ai/blog/glm-5-3-vs-glm-5-3-flash-on-deepswe-cost-coding-and-routinghttps://www.together.ai/blog/glm-5-3-vs-glm-5-3-flash-on-deepswe-cost-coding-and-routingFri, 28 Aug 2026 00:00:00 GMTblogWe ran 900 DeepSWE rollouts on GLM-5.3 and GLM-5.3 Flash. Flash gives up 5.6 points of pass@1 at 17x lower cost, and only 2.6 points at pass@4.