Architecture
data/models.yaml. Every label is auditable
against the model's sources.
Specs
- Architecture
- unknown
- Total params
- not disclosed
- Active params
- not disclosed
- Context window
- 400K tokens
- Attention
- unknown
- Position encoding
- unknown
- Post-training
- rlhf
- OSI-approved
- no
- Data released
- no
- Training code
- not released
Benchmarks
Each score carries the date it was published; we never infer or interpolate missing scores.
General reasoning
| MMLU-Pro | 87.0 | as of 2026-05-21 | source ↗ |
| GPQA-Diamond | 87.3 | as of 2026-05-21 | source ↗ |
Code
| LiveCodeBench | 86.8 | as of 2026-05-21 | source ↗ |
Math
| AIME 2025 | 94.0 | as of 2026-05-21 | source ↗ |
Available quantizations
None. The weights are not distributed, so there are no public quantizations.
Notable innovations
- · Eight selectable personality presets
- · Adaptive reasoning in Instant mode
- · GPT-5.1-Codex-Max for long-horizon agentic coding
Lineage
Tone and personality update, plus the Codex-Max agentic-coding variant.
Derived from
GPT-5 2025-08-07