Vercel AI Gateway · 2026-10-29
Vercel AI Gateway rankings
Models on the Vercel AI Gateway, ranked by the tokens they actually processed. A second, independent read on demand alongside OpenRouter.
What Vercel gateway traffic shows in October 2026
29 models carried measurable traffic on the latest archived day, totalling about — tokens across 11 labs. Other leads with 18.26% of tokens, -11.81pp over 30 days.
The "OR share" column shows each model's share on OpenRouter for the same day. Large gaps are the interesting part: a model that dominates one gateway and barely registers on the other is being pulled by a specific developer population rather than by broad demand.
Top models on the Vercel AI Gateway
Ranked by share of gateway tokens on the latest archived day.
| # | Model | Lab | Share % | 30d Δ | OR share % | $/M out |
|---|---|---|---|---|---|---|
| 1 | Other | Other | 18.26 | -11.81pp | — | — |
| 2 | Beam 1.0 | Other | 17.62 | +17.62pp | — | $1.00 |
| 3 | DeepSeek V4.1 Flash 1017 | DeepSeek | 15.19 | +15.19pp | — | $0.60 |
| 4 | DeepSeek V4.1 Flash | DeepSeek | 10.60 | -15.03pp | — | $0.60 |
| 5 | GLM 5.3 Flash | Z-AI | 10.39 | +0.14pp | — | $0.074 |
| 6 | Grok 4.7 | xAI | 6.17 | +5.79pp | — | $6.00 |
| 7 | Claude Opus 5.5 | Anthropic | 5.70 | -0.95pp | — | $20.00 |
| 8 | GLM 5.3 | Z-AI | 5.45 | +4.12pp | — | $8.80 |
| 9 | Kimi K3 | Moonshot | 4.57 | -0.81pp | — | $10.00 |
| 10 | GPT-6 Luna | OpenAI | 3.69 | -1.43pp | — | $0.50 |
| 11 | Kimi K3.1 Pro | Moonshot | 2.36 | +2.36pp | — | $10.00 |
| 12 | Claude Opus 4.8 | Anthropic | 0.00 | -1.19pp | — | $25.00 |
| 13 | Step 3.7 Flash | StepFun | 0.00 | -2.58pp | — | $1.15 |
| 14 | GLM 5.2 | Z-AI | 0.00 | -1.49pp | — | $7.00 |
| 15 | Claude Sonnet 5 | Anthropic | 0.00 | -2.36pp | — | $10.00 |
| 16 | GPT 5.6 Luna | OpenAI | 0.00 | -4.50pp | — | $1.20 |
| 17 | GPT 5.6 Sol | OpenAI | 0.00 | -1.61pp | — | $20.00 |
| 18 | Claude Opus 5 | Anthropic | 0.00 | -0.88pp | — | $20.00 |
| 19 | DeepSeek V4 Flash 0731 | DeepSeek | 0.00 | -3.68pp | — | $0.24 |
| 20 | Grok 4.5 | xAI | 0.00 | -0.02pp | — | $6.00 |
| 21 | Grok 4.6 | xAI | 0.00 | -0.38pp | — | $6.00 |
| 22 | Gemini 3.8 Flash | 0.00 | -3.77pp | — | $4.13 | |
| 23 | Muse Spark 1.3 Contributor | Other | 0.00 | -2.12pp | — | $0.20 |
| 24 | Claude Sonnet 5.5 | Anthropic | 0.00 | -0.75pp | — | $10.00 |
| 25 | GPT-6.1 Sol | OpenAI | 0.00 | -0.13pp | — | $10.00 |
| 26 | Glyph Cluster | Other | 0.00 | +0.00pp | — | — |
| 27 | Claude Haiku 5.5 | Anthropic | 0.00 | +0.00pp | — | $0.50 |
| 28 | Nemotron 3.5 Lightning Free | Nvidia | 0.00 | +0.00pp | — | Free |
| 29 | MiniMax-M3.1-Flash | MiniMax | 0.00 | +0.00pp | — | $0.60 |
Vercel AI Gateway rankings FAQ
- Which model is most used on the Vercel AI Gateway?
- Other from Other, with 18.26% of tokens routed through the Vercel AI Gateway on 2026-10-29.
- What are the top models on the Vercel AI Gateway?
- The current top three by measured token volume are Other, Beam 1.0, DeepSeek V4.1 Flash 1017, out of 29 models with traffic.
- How do Vercel AI Gateway rankings differ from OpenRouter?
- The two gateways serve different developer populations, so their mixes diverge — Vercel traffic skews toward frontend and framework-adjacent apps, OpenRouter toward a broader mix including chat frontends and open-weight experimentation. Comparing them is the fastest way to tell a genuine model trend from a single-gateway artefact.
- Where does the data come from?
- Vercel publishes model usage for its AI Gateway. Tidelines archives that snapshot once a day, so 7-day and 30-day changes compare two real archived days rather than estimates.