Loading activity…
@vipulved
Co-founder, CEO @togethercompute
AI
Together says developers can define training loops using reinforcement learning or supervised fine-tuning while it handles the infrastructure.
AI
DeepSeek says the model is live on its website, app and API, and optimized for DeepSeekHarness and Tencent CodeBuddy.
AI
finkd says the model runs well on new Mac Studios, crediting work with Apple and John Ternus.
AI
Its creators also report 10–21× faster unconditional generation on three datasets, with comparable or better generative perplexity.
AI
Reflection says Beam uses 23B active parameters and plans to publish its full weights later in October 2026.
AI
Together’s Auto router selects a model once per session; an optional Anthropic key changes the choices.
AI
Together said it ranked No. 1 in OpenRouter token share on September 29, 2026, for GLM 5.3 Flash (29.2%), DeepSeek V4.1 Flash (25.6%) and Kimi K3 (18.9%).
AI
A promotional post for Together's serverless APIs for open-weight models touted week-over-week growth and said users could get started with $5.
AI
A provider, citing OpenRouter figures, says it serves 30% of DeepSeek v4.1 requests on the platform and has the fastest throughput among supported serverless endpoints.
AI
The user claims V4.1 is cheaper to serve than V4-Flash but costs twice as much per output token even off-peak.
AI
The post claims Together serves GLM 5.3 and GLM 5.3 Flash in the top decile for tokens per second, latency and cache rate—not just one performance metric.
AI
Platform brings open models to Equinix data centers worldwide.
AI
Video creation platform behind Cinema Studio becomes a Together AI customer after its Series B.
AI
A new paper demonstrates large-scale extraction of raw reasoning traces from OpenAI, Anthropic, and Gemini models via API flaws.
AI
AI investor suggests closed labs will expose issues in open models; YouTuber replies open weights can too.
