Runware served 93.3% of input tokens from cache in September 2026, the highest among major providers. The median major provider served 75.4%.

Key findings

1. 01Runware leads at 93.3%; Vertex (Gemini) is lowest at 11.7%.
2. 028 of 17 major providers served at least 80% of input tokens from cache.
3. 03Nebius moved the most since August 2026: cache hit rate went from 75.9% to 31.8% (-44.1 pp).

Cache hit rate by month

What changed since August 2026

• Nebius75.9% → 31.8% (-44.1 pp)
• Vertex (Gemini)52.8% → 11.7% (-41.0 pp)
• OpenAI50.3% → 67.7% (+17.4 pp)
• xAI67.4% → 80.9% (+13.5 pp)
• TensorX61.7% → 73.1% (+11.3 pp)

What to do about it

• Keep the system prompt and tool definitions at the start of the request and unchanged between turns.
• On providers that need explicit cache markers, set them. Without them the hit rate stays near zero.

Data

17 rows published. Rows below the floors are withheld.

Questions this answers

Which LLM provider has the highest prompt cache hit rate?

Runware served 93.3% of input tokens from cache in September 2026, the highest among major providers. The median major provider served 75.4%.

What is a typical prompt caching hit rate in production?

Runware served 93.3% of input tokens from cache in September 2026, the highest among major providers. The median major provider served 75.4%.

How much of my input is served from cache on Anthropic, OpenAI or Bedrock?

Runware served 93.3% of input tokens from cache in September 2026, the highest among major providers. The median major provider served 75.4%.

Method and definitions

• 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.

Cache hit rate:

Cached input tokens / input tokens, summed over the month.

Per-model numbers with the same definitions are on /rankings, refreshed weekly.

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.

Revision history and other months

• October 2026first published later
• September 2026 (this page)first published 2026-11-11
• August 2026first published 2026-11-11
• July 2026first published 2026-11-11
• June 2026first published 2026-11-11
• May 2026first published 2026-11-11

Related

• Provider error ratesHow often do LLM providers return errors, and what kind?
• Time to first token by providerWhich LLM provider returns the first token fastest?
• Prompt-cache hit rate per provider, April 2026April 2026 note

Cite and reuse

Free to reuse under CC BY 4.0. Credit "Requesty Data" and link to this page. Cite the month URL so the numbers you quote stay fixed.

Requesty (2026). Prompt cache hit rate by provider, September 2026. Requesty Data. https://www.requesty.ai/data/provider-cache-hit-rate/2026-09

@misc{requesty_provider_cache_hit_rate_2026_09,
  author       = {{Requesty}},
  title        = {Prompt cache hit rate by provider, September 2026},
  year         = {2026},
  howpublished = {\url{https://www.requesty.ai/data/provider-cache-hit-rate/2026-09}},
  note         = {Requesty Data}
}

period,provider,cache_hit_rate_fraction,request_share_fraction,sample_band
2026-09,Runware,0.933,0.0096,500+
2026-09,DeepSeek,0.9049,0.0488,500+
2026-09,Novita,0.8977,0.0154,500+
…