---
id: provider-latency/2026-05
series: provider-latency
title: Time to first token by provider, May 2026
period: 2026-05
period_start: 2026-05-01
period_end: 2026-05-31
updated: 2026-11-11
license: CC BY 4.0
canonical: https://www.requesty.ai/data/provider-latency/2026-05
latest: https://www.requesty.ai/data/provider-latency
---

# Time to first token by provider, May 2026

> Azure had the fastest median time to first token among major providers in May 2026 at 548 ms. The median major provider took 1.24 s.

Question: Which LLM provider returns the first token fastest?

## Key findings

1. Fastest median TTFT: Azure at 548 ms, then Nebius at 819 ms.
2. Highest median output speed: Vertex (Gemini) at 141 tokens per second.
3. Fastest median full streamed response: Azure at 980 ms.

## Data

| Provider | Median TTFT | p90 TTFT | Median total | Median tokens/s | Streamed share | Organizations (band) |
| --- | --- | --- | --- | --- | --- | --- |
| Azure | 548 ms | 880 ms | 980 ms | 85.700 | 79.83% | 100-499 |
| Nebius | 819 ms | 1.81 s | 3.49 s | 130.000 | 4.64% | 100-499 |
| DeepSeek | 875 ms | 1.67 s | 5.40 s | 44.000 | 12.60% | 100-499 |
| xAI | 899 ms | 8.03 s | 8.00 s | 25.900 | 4.82% | 100-499 |
| Vertex (Claude) | 1.14 s | 2.72 s | 5.86 s | 73.300 | 35.58% | 100-499 |
| DeepInfra | 1.24 s | 5.75 s | 6.82 s | 49.600 | 1.46% | 25-99 |
| Anthropic | 2.02 s | 3.33 s | 5.53 s | 80.900 | 49.87% | 500+ |
| OpenAI | 2.33 s | 16.7 s | 6.03 s | 85.900 | 5.67% | 500+ |
| Bedrock | 2.39 s | 5.91 s | 6.14 s | 89.200 | 17.70% | 100-499 |
| Novita | 2.40 s | 4.81 s | 5.82 s | 35.900 | 0.76% | 100-499 |
| Vertex (Gemini) | 2.50 s | 16.6 s | 3.48 s | 141.400 | 3.36% | 500+ |

## Method

- Time to first token and tokens per second are measured on streamed, successful requests only, with the same formulas /rankings uses. Tokens per second counts visible output tokens after the first token, with a 250 ms floor on decode time and at least 20 output tokens.
- A provider needs at least 500 streamed successes in the month for latency to be published. Only providers with at least 0.5% of published traffic are shown.
- A row is published only when it has at least 20 requests, at least 5 distinct organizations, and no single organization above 80% of its traffic. These are the same floors /rankings uses. Rows below the floors are withheld, not estimated.
- Only shares, rates, medians and organization bands are published. Absolute request, token and spend totals stay private.

## Caveats

- Latency depends on the model mix. A provider serving mostly small models will look faster than one serving large reasoning models.
- Reasoning models think before the first visible token, which raises TTFT.

## Cite as

Requesty (2026). Time to first token by provider, May 2026. Requesty Data. https://www.requesty.ai/data/provider-latency/2026-05

```bibtex
@misc{requesty_provider_latency_2026_05,
  author       = {{Requesty}},
  title        = {Time to first token by provider, May 2026},
  year         = {2026},
  howpublished = {\url{https://www.requesty.ai/data/provider-latency/2026-05}},
  note         = {Requesty Data}
}
```

Source: Requesty production gateway. License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).
Downloads: [JSON](https://www.requesty.ai/data/provider-latency/2026-05/data.json) · [CSV](https://www.requesty.ai/data/provider-latency/2026-05/data.csv) · [Markdown](https://www.requesty.ai/data/provider-latency/2026-05/data.md)