Agentic Text Synthesis on RAVEL
95.7Success Rate (S%)DeepSeek-v3.2
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek-v3.2Model Category=Proprietary LLMs2026.02 | 95.7 | 2.59 | 26.2 | 2.18 | 6.64 | 0.02 | 7 | 17.9 | |
| Gemini-3 ProModel Category=Proprietary LLMs2026.02 | 95.1 | 2.32 | 2.4 | 27.96 | 7.05 | 0 | 7.1 | 16.2 | |
| Qwen3-8BModel Category=Open Source LLMs2026.02 | 94.7 | 2.43 | 13.9 | 2.57 | 5.51 | 0.02 | 7.5 | 18 | |
| Hermes-3-Llama-70BModel Category=Open Source LLMs2026.02 | 91.8 | 3.16 | 18.7 | 11.39 | 4.58 | 0 | 7.7 | 24.1 | |
| Qwen3-235B-A22BModel Category=Open Source LLMs2026.02 | 89.4 | 2.47 | 20.5 | 0 | 6.94 | 0 | 8.3 | 20.1 | |
| GLM-4.7Model Category=Proprietary LLMs2026.02 | 73.3 | 2.51 | 38.5 | 0 | 6.31 | 0.37 | 7.3 | 18 | |
| Grok-4Model Category=Proprietary LLMs2026.02 | 69 | 2.5 | 38.2 | 3.52 | 6.31 | 0.03 | 7.4 | 18.4 | |
| Claude-4.5 SonnetModel Category=Proprietary LLMs2026.02 | 68.1 | 2.95 | 100.5 | 19.55 | 6.68 | 0 | 10.3 | 28 | |
| GPT-5.2-2025-11-20Model Category=Proprietary LLMs2026.02 | 64.3 | 2.35 | 17.9 | 0 | 6.64 | 0.95 | 15.5 | 30.2 | |
| Qwen3-32BModel Category=Open Source LLMs2026.02 | 58.5 | 3.02 | 105.8 | 5.54 | 5.54 | 0 | 7.2 | 21.2 | |
| Kimi-K2-ThinkingModel Category=Proprietary LLMs2026.02 | 58.3 | 2.81 | 55.4 | 0 | 6.48 | 1.32 | 7.9 | 21.6 | |
| Qwen3-MaxModel Category=Proprietary LLMs2026.02 | 47.8 | 1.92 | 23.9 | 0 | 5.74 | 0 | 7.3 | 13.1 | |
| Llama-3.1-405BModel Category=Open Source LLMs2026.02 | 47.7 | 3.94 | 30.3 | 1.09 | 5.05 | 0 | 7.5 | 29.6 | |
| Llama-3.1-8BModel Category=Open Source LLMs2026.02 | 0 | 8.59 | 18.3 | 4.6 | 1.66 | 0.04 | 5.8 | 51 |