Most-used AI models for file i/o
DeepSeek V4 Flash 0423 is the most-used model for file i/o on OpenRouter, handling 11.5% of requests classified as file i/o over the last 7 days.
This ranks models by how often they are actually chosen for file i/o in production traffic, not by benchmark score. Adoption reflects price, latency and availability as much as raw capability — which is exactly why it often disagrees with the benchmark leaderboards, and why it is worth reading alongside them.
| # | Model | Request share | Token share | Input /M | Context |
|---|---|---|---|---|---|
| 1 |  DeepSeek V4 Flash 0423 | 11.5% | 10.0% | $0.140 | 1.0M |
| 2 |  Hy3 | 10.8% | 12.9% | $0.132 | 262K |
| 3 |  MiMo-V2.5 | 10.6% | 11.3% | $0.140 | 1.1M |
| 4 |  Step 3.7 Flash | 7.4% | 6.1% | $0.200 | 262K |
| 5 | DeepSeek V4 Pro | 6.3% | 4.8% | $0.435 | 1.0M |
| 6 | DeepSeek V4 Flash 0423 | 5.6% | 8.3% | $0.140 | 1.0M |
| 7 |  GLM 5.2 | 5.3% | 5.5% | $0.760 | 1.0M |
| 8 |  GPT-5.6 Luna | 4.4% | 6.0% | $0.100 | 1.1M |
| 9 | nemotron-3-ultra-550b-a55b-20260604:free | 3.1% | 4.9% | — | — |
| 10 |  Claude Sonnet 4.6 | 3.0% | 1.3% | $3.00 | 1M |
Reading the two columns together: nemotron-3-ultra-550b-a55b-20260604:free takes a noticeably larger share of tokens than of requests — meaning it is being used for the longer, heavier file i/o jobs rather than quick one-shot calls.
Frequently asked
DeepSeek V4 Flash 0423 is the most-used model for file i/o on OpenRouter, handling 11.5% of requests classified as file i/o over the last 7 days. It is followed by Hy3 (10.8%) and MiMo-V2.5 (10.6%).
No. This ranks by how often each model is actually chosen for file i/o in production traffic through OpenRouter — real-world adoption, which reflects price and availability as much as capability. For capability-based rankings, see the benchmark leaderboards.
File I/O accounts for 1.4% of classified requests and 4.7% of classified tokens on OpenRouter over the trailing 7 days.
Other tasks
Source: OpenRouter (openrouter.ai/rankings), as of 2026-08-04. Shares are of classified, sampled traffic over a trailing 7-day window; absolute volumes are not published.