Stepfun
Step 5 Preview
Launched 2026-09-16 · Listed on OpenRouter 2026-10-08
Prices updated . Benchmarks updated .
Step 5 Preview is StepFun's flagship model for agentic work, built on a sparse Mixture-of-Experts architecture (27B active / 600B total parameters). It performs strongly in software engineering and professional...
You send / 1M
$1.00
tokens in
It replies / 1M
$2.70
tokens out
Context
1M
tokens it can read at once
Typical request
$0.0027
1,000 words in, 400 out
Benchmarks
| Benchmark | Score | Kind | Date | Source |
|---|---|---|---|---|
| CritPt | 20.9% | Independent | 2026-09-18 | CritPt, republished by Epoch AI |
| Arena Elo (text) | 1,443 | Independent | 2026-10-28 | LMArena |
| Arena Elo (webdev)High effort | 1,571 | Independent | 2026-10-28 | LMArena |
| Arena Elo (vision) | 1,265 | Independent | 2026-10-19 | LMArena |
Data privacy
No training or retention: Yes. At least one provider serves it with zero data retention.
| Provider | You send / 1M | It replies / 1M | Uptime | Privacy policy |
|---|---|---|---|---|
| Featherless · fp4 | $1.00 | $2.70 | 38% | Read |
| StepFun · fp8 | $1.00 | $2.70 | 100% | n/a |
Based on OpenRouter and Vercel AI Gateway data. Confirm terms with the provider before sending sensitive data.
What it can do
- Reads: text, image, video
- Writes: text
- Weights: Open weights
- Reasoning: Yes
- Tool calling: Yes
- Longest reply: 64K tokens
- Repeated text is billed at $0.100 per 1M (prompt caching)
Compare with
Appears in
Top No-Training AI ModelsTop No-Training Zero Data Retention AI ModelsTop No-Training Zero Data Retention Vision AI ModelsTop No-Training Zero Data Retention Reasoning AI ModelsTop No-Training Zero Data Retention Tool-Calling AI ModelsTop No-Training Zero Data Retention Long-Context AI ModelsTop No-Training Vision AI ModelsTop No-Training Vision Reasoning AI Models
Estimate your monthly cost with Step 5 Preview
Sources and licenses
- OpenRouter: Prices, context windows, modalities and the model list come from the OpenRouter public API. Data: OpenRouter.
- Vercel AI Gateway: The no-training flag comes from the Vercel AI Gateway model list.
- models.dev: Lab launch dates, open-weights flags and knowledge cutoffs come from models.dev. Copyright (c) 2025 models.dev, MIT License.
- Epoch AI: Independent benchmark scores and the Epoch Capabilities Index (ECI) come from Epoch AI's Benchmarking Hub, used under CC BY 4.0. Epoch AI, Capabilities & benchmarking, epoch.ai/benchmarks. Scores Epoch marks as external keep their original license, and the source is named on each score.
- LMArena: Arena Elo (text, webdev and vision) comes from the LMArena leaderboard dataset (lmarena-ai/leaderboard-dataset), CC BY 4.0, leaderboard date 2026-10-28.
- Lab-reported scores, when an editor has added them, are the lab's own numbers. Each one links to the source and is labeled lab-reported. They are not used to rank the lists.
- What a score measures: the benchmark guide.
