---
id: provider-cache-hit-rate/2026-10
series: provider-cache-hit-rate
title: Prompt cache hit rate by provider, October 2026
period: 2026-10
period_start: 2026-10-01
period_end: 2026-10-31
updated: 2026-11-11
license: CC BY 4.0
canonical: https://www.requesty.ai/data/provider-cache-hit-rate/2026-10
latest: https://www.requesty.ai/data/provider-cache-hit-rate
---

# Prompt cache hit rate by provider, October 2026

> Fireworks served 94.0% of input tokens from cache in October 2026, the highest among major providers. The median major provider served 80.4%.

Question: What share of input tokens do LLM providers serve from prompt cache?

## Key findings

1. Fireworks leads at 94.0%; Vertex (Gemini) is lowest at 51.1%.
2. 9 of 17 major providers served at least 80% of input tokens from cache.
3. Vertex (Gemini) moved the most since September 2026: cache hit rate went from 11.7% to 51.1% (+39.3 pp).

## Data

| Provider | Cache hit rate | Traffic share | Organizations (band) |
| --- | --- | --- | --- |
| Fireworks | 94.04% | 0.84% | 500+ |
| Melious | 90.10% | 1.36% | 500+ |
| DeepSeek | 88.53% | 3.41% | 500+ |
| Novita | 84.53% | 1.48% | 500+ |
| Bedrock | 82.41% | 7.74% | 500+ |
| Vertex (Claude) | 82.28% | 2.08% | 500+ |
| Sference | 82.23% | 7.24% | 500+ |
| NVIDIA | 80.75% | 0.84% | 500+ |
| Anthropic | 80.36% | 3.47% | 500+ |
| Lyceum | 79.31% | 0.85% | 500+ |
| OpenAI | 77.73% | 11.25% | 500+ |
| DeepInfra | 75.98% | 11.37% | 500+ |
| Azure | 72.39% | 12.99% | 500+ |
| xAI | 72.20% | 0.59% | 100-499 |
| TensorX | 66.34% | 6.51% | 500+ |
| Mistral | 61.30% | 1.03% | 500+ |
| Vertex (Gemini) | 51.06% | 11.09% | 500+ |

## Method

- Cache hit rate is cached input tokens divided by input tokens, the same definition /rankings uses.
- Providers that report no cached tokens are left out of the chart. 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

- Hit rate depends on workload. Long agent sessions with a stable prefix cache far better than one-off requests.
- Providers report cached tokens differently. Some only count explicit cache reads.

## Cite as

Requesty (2026). Prompt cache hit rate by provider, October 2026. Requesty Data. https://www.requesty.ai/data/provider-cache-hit-rate/2026-10

```bibtex
@misc{requesty_provider_cache_hit_rate_2026_10,
  author       = {{Requesty}},
  title        = {Prompt cache hit rate by provider, October 2026},
  year         = {2026},
  howpublished = {\url{https://www.requesty.ai/data/provider-cache-hit-rate/2026-10}},
  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-cache-hit-rate/2026-10/data.json) · [CSV](https://www.requesty.ai/data/provider-cache-hit-rate/2026-10/data.csv) · [Markdown](https://www.requesty.ai/data/provider-cache-hit-rate/2026-10/data.md)