Updated daily at 05:00 UTC
- Released
- Jan 2026
- Status
- Legacy
- Context
- 262.1K
- Max output
- —
- Input per 1M tokens
- $0.10
- Output per 1M tokens
- $0.30
- Open weights
- Yes
- License
- apache-2.0
- Parameters
- 196B, 11B active
- Knowledge cutoff
- —
- Reasoning
- Yes
- Modalities
- text → text
Category scores
Out of 100, from independent and third-party results only, by method category-2026-09-29.1.
| Category | Score | Rank | Coverage |
|---|---|---|---|
| Human preference | 49.0 | Not ranked | 1 of 5 |
Benchmark results
Every result we hold for this model, with who ran it and where it was published.
Knowledge & reasoning
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Humanity's Last Exam (HLE) | Accuracy | 23.1% | Self-reported | — | — | Step 3.5 Flash Paper via Hugging Face Community Evals |
| MMLU-Pro | Accuracy | 84.4% | Self-reported | — | — | Step 3.5 Flash Paper via Hugging Face Community Evals |
Scientific reasoning
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| GPQA Diamond | Accuracy | 83.5% | Self-reported | — | — | Step 3.5 Flash Paper via Hugging Face Community Evals |
Human preference
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Text Arena (arena.ai) | Arena score (Elo scale) | 1393 | Independent | — | 25 Sep 2026 | Arena |
| Text Arena (arena.ai) | Votes | 60243 | Independent | — | 25 Sep 2026 | Arena |
Other Step versions
- Step 5 PreviewSep 2026
- Step 3.7 FlashMay 2026
- Step3Jul 2025
- Step-2Jul 2024
- Step-1—
- Step-2-mini—