Grok-4.20-beta-0309-non-reasoning is the ultra-fast, efficiency-first variant of the Grok-4.20 family. Released as part of the March 2026 update, this model bypasses the "Reasoning Trace" (Chain-of-Thought) layers to provide near-instantaneous responses for high-volume data processing. It is powered by the Colossus supercluster and features a "Rapid Learning" architecture that incorporates real-time X platform data with millisecond latency. This model is ideal for tasks where immediate pattern recognition, fast drafting, or real-time information retrieval from X (Twitter) is required, offering the lowest hallucination rate in the non-reasoning class through its "Society of Mind" multi-agent verification (performed silently in the background).
所有提供方:為該模型提供服務的每個上游都支援。部分提供方:取決於處理請求的上游。預設值:未設定時傳送的值。
| 參數 | 提供方 | 預設值 |
|---|---|---|
| frequency_penalty | 部分提供方 | 預設不傳送 |
| include_reasoning | 所有提供方 | - |
| logprobs | 所有提供方 | - |
| max_tokens | 所有提供方 | - |
| presence_penalty | 部分提供方 | 預設不傳送 |
| reasoning | 所有提供方 | - |
| repetition_penalty | 部分提供方 | 預設不傳送 |
| response_format | 所有提供方 | - |
| seed | 所有提供方 | - |
| structured_outputs | 所有提供方 | - |
| temperature | 所有提供方 | 預設不傳送 |
| tool_choice | 所有提供方 | - |
| tools | 所有提供方 | - |
| top_k | 部分提供方 | 預設不傳送 |
| top_logprobs | 所有提供方 | - |
| top_p | 所有提供方 | 預設不傳送 |