RinkoAI provides OpenAI-compatible API endpoints and other model interfaces with a focus on stability and cost-effectiveness.
| Model | Speed | Latency | Tests |
|---|---|---|---|
| gemini-2.0-flash | 12473.78 t/s | 6.15s | 5 |
| Qwen/Qwen3-Coder-480B | 552.48 t/s | 0.89s | 5 |
| Qwen/Qwen3-Coder-480B | 552.48 t/s | 0.89s | 5 |
| gpt-oss-120b | 335.21 t/s | 1.40s | 35 |
| Qwen/Qwen3-235B | 272.18 t/s | 1.71s | 25 |
| moonshotai/kimi-k2-instruct-0905 | 217.99 t/s | 0.88s | 40 |
| deepseek-ai/DeepSeek-V3.1 | 102.83 t/s | 1.76s | 10 |
| deepseek-ai/DeepSeek-V3.1 | 102.83 t/s | 1.76s | 10 |
| gpt-4.1-mini | 79.59 t/s | 2.60s | 10 |
| Qwen/Qwen3-30B-A3B | 73.72 t/s | 1.37s | 10 |
| xai/grok-4-fast-non-reasoning | 69.29 t/s | 1.87s | 25 |
| claude-4.0-sonnet-search | 52.61 t/s | 4.12s | 5 |
| MistralAI/Mistral-Nemo | 40.81 t/s | 2.15s | 10 |
| MistralAI/Mistral-Large | 39.11 t/s | 2.12s | 5 |
| Qwen/Qwen3-14B | 38.30 t/s | 1.19s | 5 |
| deepseek-ai/DeepSeek-V3.2-Exp | 30.90 t/s | 2.35s | 5 |
| deepseek-ai/DeepSeek-V3.2-Exp | 30.90 t/s | 2.35s | 5 |
| Time | Model | Speed | Latency |
|---|---|---|---|
| Dec 23, 03:04 AM | Unknown | - | -s |
| Dec 21, 01:49 PM | Unknown | - | -s |
| Dec 21, 01:43 PM | Unknown | - | -s |
| Nov 20, 03:31 AM | gpt-4.1-mini | 78.99 t/s | 1.62s |
| Nov 9, 09:35 AM | gpt-oss-120b | 590.03 t/s | 0.84s |
| Nov 9, 09:33 AM | moonshotai/kimi-k2-instruct-0905 | 227.97 t/s | 0.78s |
| Oct 26, 02:49 PM | Qwen/Qwen3-235B | 76.50 t/s | 1.69s |
| Oct 26, 02:47 PM | gpt-4.1-mini | 80.19 t/s | 3.57s |
| Oct 26, 11:36 AM | moonshotai/kimi-k2-instruct-0905 | 224.98 t/s | 0.89s |
| Oct 26, 11:35 AM | MistralAI/Mistral-Nemo | 42.93 t/s | 2.50s |