Start from zero. Stay up to date. Build solutions yourself—or leave it to us.
5.2GLMAn MIT-licensed, open-weight Mixture-of-Experts model built for long-horizon agentic engineering — near-frontier coding and multi-hour autonomous tool use at a fraction of closed-model cost, shadowed by brittle tool-call parsing and heavy self-hosting requirements.
5 FableClaudeThe first public release of Anthropic's top-tier Mythos system — suspended three days after launch by a US government export-control directive (the first deployed AI model taken offline by government order), restored three weeks later, and now the author of its own report. API-only after July 7, 2026.
4.8 OpusClaudeAnthropic's frontier reasoning model: a deliberate, self-auditing problem-solver tuned for hard, multi-step work where being right matters more than being fast.
