Update xAI model references for Grok model retirement (#1667)

grok-2-1212 and grok-3 are being retired by xAI on May 15, 2026.
Update model reference table and example config to use grok-4.3.

Author: Jeadie

website/docs/reference/models.md

@@ -20,7 +20,7 @@ For more details on how model grades are evaluated in Spice, refer to the [model
 | `o1-2024-12-17 (Reasoning effort: low)`         | **C**       | `openai`       | 200k tokens<br/ >100k tokens          | ✅              | ✅                                       | ✅    | ✅                          | ✅        | ❌        | 2024-12-17         | v1.0.2        |
 | `gpt-4o-2024-08-06`                             | **B**       | `openai`       | 128k tokens<br/ >16384 tokens         | ✅              | ✅                                       | ✅    | ✅                          | ❌        | ✅        | 2024-08-06         | v1.0.2        |
 | `claude-3-5-sonnet-20241022`                    | **C**       | `anthropic`    | 200k tokens<br/ >8192 tokens          | ✅              | ❌                                       | ✅    | ✅                          | ❌        | ✅        | 2024-10-22         | v1.0.2        |
-| `grok-2-1212`                                   | Ungraded    | `xai`          | −                                     | ✅              | −                                        | −     | −                           | ❌        | −         | Not Available      | v1.0.2        |
+| `grok-4.3`                                      | Ungraded    | `xai`          | 1M tokens<br/ >−                      | ✅              | −                                        | −     | −                           | ✅        | −         | 2026               | v2.1          |
 | `deepseek-ai/DeepSeek-R1-Distill-Llama-8B`      | Ungraded    | `huggingface`  | −                                     | ✅              | −                                        | −     | −                           | ✅        | −         | Not Available      | v1.0.2        |
 | `meta-llama/Llama-3.2-3B-Instruct`              | Ungraded    | `huggingface`  | −                                     | ✅              | −                                        | −     | −                           | ❌        | −         | Not Available      | v1.0.2        |
 | `deepseek-reasoner`                             | Ungraded    | `openai`       | 64k tokens<br/ >8k tokens             | ✅              | −                                        | −     | −                           | ✅        | ✅        | 2025/01/20         | v1.0.4        |