---
id: provider-latency/2026-06
series: provider-latency
title: Time to first token by provider, June 2026
period: 2026-06
period_start: 2026-06-01
period_end: 2026-06-30
updated: 2026-11-11
license: CC BY 4.0
canonical: https://www.requesty.ai/data/provider-latency/2026-06
latest: https://www.requesty.ai/data/provider-latency
---

# Time to first token by provider, June 2026

> Azure had the fastest median time to first token among major providers in June 2026 at 618 ms. The median major provider took 1.49 s.

Question: Which LLM provider returns the first token fastest?

## Key findings

1. Fastest median TTFT: Azure at 618 ms, then DeepSeek at 847 ms.
2. Highest median output speed: Vertex (Gemini) at 154 tokens per second.
3. Fastest median full streamed response: Azure at 1.15 s.
4. OpenAI moved the most since May 2026: median TTFT went from 2.33 s to 3.48 s (+1.15 s).

## Data

| Provider | Median TTFT | p90 TTFT | Median total | Median tokens/s | Streamed share | Organizations (band) |
| --- | --- | --- | --- | --- | --- | --- |
| Azure | 618 ms | 1.61 s | 1.15 s | 90.400 | 73.26% | 100-499 |
| DeepSeek | 847 ms | 1.50 s | 5.17 s | 46.500 | 14.72% | 100-499 |
| Vertex (Claude) | 1.25 s | 2.99 s | 4.76 s | 88.000 | 71.03% | 100-499 |
| xAI | 1.29 s | 6.73 s | 10.9 s | 30.200 | 9.09% | 100-499 |
| Mistral | 1.35 s | 8.69 s | 3.36 s | 76.400 | 16.87% | 100-499 |
| Fireworks | 1.36 s | 6.91 s | 5.04 s | 112.200 | 82.50% | 100-499 |
| DeepInfra | 1.62 s | 6.73 s | 8.86 s | 26.800 | 14.49% | 100-499 |
| Anthropic | 1.96 s | 3.64 s | 6.24 s | 80.500 | 55.09% | 500+ |
| Novita | 1.97 s | 4.93 s | 8.69 s | 42.400 | 4.06% | 100-499 |
| Bedrock | 2.07 s | 6.43 s | 5.92 s | 85.300 | 33.48% | 500+ |
| Vertex (Gemini) | 2.63 s | 15.0 s | 3.39 s | 154.300 | 7.02% | 500+ |
| OpenAI | 3.48 s | 15.3 s | 5.82 s | 104.100 | 8.21% | 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, June 2026. Requesty Data. https://www.requesty.ai/data/provider-latency/2026-06

```bibtex
@misc{requesty_provider_latency_2026_06,
  author       = {{Requesty}},
  title        = {Time to first token by provider, June 2026},
  year         = {2026},
  howpublished = {\url{https://www.requesty.ai/data/provider-latency/2026-06}},
  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/2026-06/data.json) · [CSV](https://www.requesty.ai/data/provider-latency/2026-06/data.csv) · [Markdown](https://www.requesty.ai/data/provider-latency/2026-06/data.md)