Research

Our latest work in AI — models, benchmarks, and the thinking behind Guardian.

Introducing G-2.0-Code

The second-generation agentic coding model from OQENYX. BFCL-V4 72, SWE-bench Verified 77% — built to fix bugs and ship code in real production codebases.

Mar 31, 2026·3 min read

Read article

Model ReleaseBenchmarks

Introducing G-2.0-Lite

The second-generation compact model from OQENYX. AIME 93, MATH-500 97, native vision — frontier-class reasoning in a fast, efficient package.

Mar 31, 2026·2 min read

Read article

Model ReleaseBenchmarks

Introducing Guardian 2.0 Thinking

The second generation of OQENYX's flagship reasoning model. IFBench 76.5, GPQA Diamond 86, MMLU-Pro 85 — and native support for 200+ languages.

Mar 31, 2026·3 min read

Read article

Model ReleaseBenchmarks

Introducing Guardian Pulse 2.0

OQENYX's multimodal model. Speech recognition in 113 languages, text-to-speech in 36 — audio, video, and text unified in one Guardian endpoint.

Mar 31, 2026·3 min read

Read article

Model ReleaseBenchmarks

Introducing Guardian 1.0 Thinking

Our first reasoning model — built for privacy, precision, and real-world tasks. Here's how Guardian 1.0 Thinking performs against leading AI models.

Mar 25, 2026·3 min read

Read article