Most-used AI models for code generation
MiMo-V2.5 is the most-used model for code generation on OpenRouter, handling 8.7% of requests classified as code generation over the last 7 days.
This ranks models by how often they are actually chosen for code generation 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 |  MiMo-V2.5 | 8.7% | 11.8% | $0.140 | 1.1M |
| 2 |  DeepSeek V4 Flash 0423 | 8.7% | 7.9% | $0.140 | 1.0M |
| 3 |  GLM 5.2 | 7.5% | 7.5% | $0.760 | 1.0M |
| 4 | DeepSeek V4 Pro | 7.0% | 5.1% | $0.435 | 1.0M |
| 5 | DeepSeek V4 Flash 0423 | 6.2% | 9.6% | $0.140 | 1.0M |
| 6 |  GPT-5.6 Luna | 5.8% | 8.1% | $0.100 | 1.1M |
| 7 |  Claude Sonnet 4.6 | 4.3% | 2.4% | $3.00 | 1M |
| 8 |  Hy3 | 4.0% | 4.3% | $0.132 | 262K |
| 9 | Claude Opus 5 | 3.4% | 2.6% | $5.00 | 1M |
| 10 |  Step 3.7 Flash | 3.4% | 4.2% | $0.200 | 262K |
Reading the two columns together: DeepSeek V4 Flash 0423 takes a noticeably larger share of tokens than of requests — meaning it is being used for the longer, heavier code generation jobs rather than quick one-shot calls.
Frequently asked
MiMo-V2.5 is the most-used model for code generation on OpenRouter, handling 8.7% of requests classified as code generation over the last 7 days. It is followed by DeepSeek V4 Flash 0423 (8.7%) and GLM 5.2 (7.5%).
No. This ranks by how often each model is actually chosen for code generation 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.
Code Generation accounts for 2.4% of classified requests and 8.9% 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.