New models are now available in Atoms🚀
Claude Opus 5
We’ve added Claude Opus 5 to Atoms. Built for long-horizon work, with stronger planning, richer outputs, and better recovery across complex builds.
DeepSeek V4 Pro & Flash
We’ve added DeepSeek V4 Pro and Flash to Atoms. Get 1M context, with Pro for complex reasoning and Flash for fast execution—estimated at 6% of Opus 5 for Flash in our internal benchmark and 18% for Pro under the same usage profile, based on current pricing.
Gemini 3.7 Flash
We’ve added Gemini 3.7 Flash to Atoms. Google’s most intelligent workhorse model, built for fast, capable performance across everyday tasks.
Qwen3.8-Max
We’ve added Qwen3.8-Max to Atoms. Qwen’s most capable model yet, built for demanding coding, research, and collaborative work.