---
id: tool-call-token-share-october-2026
slug: tool-call-token-share-october-2026
title: "Tool-call share of output tokens and requests, October 2026"
topic: agentic
period: Oct 2026
updated: 2026-11-11
license: CC BY 4.0
canonical: https://www.requesty.ai/data/tool-call-token-share-october-2026
---

# Tool-call share of output tokens and requests, October 2026

> How much provider traffic is tool calling? In October 2026 Melious had the highest tool-call share at 66.4% of output tokens and 66.2% of requests. Across the busiest providers, tool calls made up between 7.2% and 66.2% of successful requests.

*Topic: Agentic workloads. Period: Oct 2026. Last updated 2026-11-11.*

## Why it matters

Tool calls are the core loop of agents. Comparing token share with request share shows whether tool-calling turns are longer or shorter than the average response on that provider.

## Questions this answers

- Which providers handle the most tool calls?
- What share of LLM output is tool calling?
- Are tool-call responses longer than other responses?

## Key findings

1. Melious: 66.4% of tokens, 66.2% of requests.
2. Lyceum: 52.4% of tokens, 50.6% of requests.
3. Vertex (Claude): 46.6% of tokens, 36.0% of requests.
4. Novita: 42.2% of tokens, 9.4% of requests.

## Data

| Provider | Tool-call output-token share (percent) | Tool-call request share (percent) | Gap (token - request) (percent) |
| --- | --- | --- | --- |
| Vertex (Gemini) | 23.50% | 18.30% | 5.20% |
| Azure | 31.70% | 31.50% | 0.20% |
| TensorX | 18.00% | 13.80% | 4.30% |
| OpenAI | 21.40% | 13.70% | 7.70% |
| DeepInfra | 3.10% | 10.40% | -7.40% |
| Nebius | 14.00% | 7.20% | 6.70% |
| Bedrock | 26.10% | 20.20% | 5.80% |
| Sference | 38.00% | 38.00% | 0.00% |
| Anthropic | 23.40% | 22.70% | 0.70% |
| DeepSeek | 18.90% | 31.80% | -12.90% |
| Vertex (Claude) | 46.60% | 36.00% | 10.60% |
| Novita | 42.20% | 9.40% | 32.80% |
| Melious | 66.40% | 66.20% | 0.20% |
| Mistral | 24.30% | 26.10% | -1.80% |
| Lyceum | 52.40% | 50.60% | 1.80% |
| Fireworks | 32.60% | 40.70% | -8.10% |

## Caveats

- Successful requests with output tokens; a request counts as a tool call when finish_reason is tool_calls.

## Cite as

**APA.** Requesty (2026). Tool-call share of output tokens and requests, October 2026. Requesty Data. https://www.requesty.ai/data/tool-call-token-share-october-2026

```bibtex
@misc{requesty_tool_call_token_share_october_2026,
  author       = {{Requesty}},
  title        = {Tool-call share of output tokens and requests, October 2026},
  year         = {2026},
  howpublished = {\url{https://www.requesty.ai/data/tool-call-token-share-october-2026}},
  note         = {Requesty Data}
}
```

---

Downloads: [JSON](https://www.requesty.ai/data/tool-call-token-share-october-2026/data.json) · [CSV](https://www.requesty.ai/data/tool-call-token-share-october-2026/data.csv) · [Markdown](https://www.requesty.ai/data/tool-call-token-share-october-2026/data.md)