xAI
Grok 4.7
最大上下文
500K
最大輸出
500K
發布日期
暫無
grok-4.5Grok 4.5 是 xAI 針對程式設計、智能體任務與知識工作的推理模型,支援文字和圖像輸入、最高 50 萬 token 上下文,以及低/中/高可配置推理強度。它支援函式呼叫、網頁與 X 搜尋、程式碼執行和結構化輸出,適合長上下文分析、多步驟工具呼叫與端到端工程工作流程;不應僅被定位為與 X 平台關聯的聊天機器人。
上下文視窗
500K Token
最大輸出
500K Token
發布日期
2026年7月8日
模態
| Token 階梯 | 輸入價格 | 輸出價格 | 快取讀取 |
|---|---|---|---|
| <=200K | $2/M | $6/M | $0.3/M |
| >200K | $4/M | $12/M | $0.6/M |
推理
工具呼叫
Temperature 參數
附件
知識庫
Endpoint 協議
Grok 4.5 結合程式設計導向推理、Agent 任務執行、圖片理解與大型內容脈絡視窗,適合軟體與工程工作。
模型面向需要規劃、撰寫程式碼與協調執行多個工程步驟的軟體工作流程。
推理能力支援對軟體、工程、科學與數學問題進行多步驟分析。
模型可在 500,000 token 內容脈絡視窗中分析文字與圖片,容納大量程式碼與專案視覺資料。
Grok 4.5 適合程式碼儲存庫開發、技術問題排查,以及結合螢幕截圖或圖表的實作任務。
分析既有程式碼庫、規劃跨檔案修改,並在遵循專案慣例的前提下實作完整功能。
結合程式碼與診斷輸出追蹤故障、評估可能原因,並制定包含驗證步驟的精準修正方案。
將螢幕截圖、介面參考或圖表轉化為具體程式碼修改,並診斷預期成果與實際呈現之間的差異。
import OpenAI from "openai"
const client = new OpenAI({
apiKey: process.env.TOKENHUB_API_KEY,
baseURL: "https://us-api.tokenhub.com/v1",
})
const result = await client.chat.completions.create({})
console.log(result.choices[0]?.message?.content)Grok 4.5 (high)
| 指數分數 | ||
|---|---|---|
| Artificial Analysis 綜合能力指數 | Artificial Analysis 綜合能力評分 | 53.8 |
| Artificial Analysis 程式碼能力指數 | Artificial Analysis 軟體任務評分 | 72.4 |
| 知識與推理 | ||
| GPQA | 高階科學問題求解 | 93.1% |
| HLE | 廣泛專家級考試集合 | 40.3% |
| 程式碼與工程 | ||
| SciCode | 科學計算程式碼挑戰 | 54.1% |
| 指令遵循與 Agent 任務 | ||
| AA-LCR | 長上下文推理 | 67.7% |
指標來源 Artificial Analysis
在 TokenHub 使用 Grok 4.5 的常見問題。
Grok 4.5 是一款面向程式設計、代理任務和知識工作的前沿 AI 模型。
它適合軟體工程、多步驟工具工作流程、程式碼審查、除錯和技術分析。
其已公開的優勢包括程式設計、代理推理、可設定的推理強度,以及函式呼叫和程式碼執行等工具支援。
模型輸出仍可能出錯,尤其是在長流程或高影響任務中。請在自身任務上測試提示詞,審查重要程式碼,並與其他模型比較延遲和成本。
在 TokenHub API 請求中將 grok-4.5 設為模型,然後使用帳戶文件中提供的介面位址、驗證方式和請求格式。
提供方文件列出的價格為每百萬輸入權杖 2 美元、每百萬輸出權杖 6 美元;TokenHub 的可用性和計費可能不同。若程式設計或工具驅動任務是重點,可優先考慮該模型,並在大規模使用前結合實際工作負載驗證。
透過一個 API Key 接入 Grok 4.5 與更多 AI 模型。
Grok 4.5 媒體與討論
精選的 Grok 4.5 公開 X 貼文、YouTube 影片和 Reddit 討論。
X (Twitter)
Eric Jiang ·
Grok 4.5 is my default for all my coding work, I freaking love this model. Frontier intelligence that won't break the bank. Grok 4.5: $2 in / $6 out GPT 5.6: $5 in / $30 out Opus 4.8: $5 in / $25 out
查看原文BridgeMind ·
New CursorBench results just dropped and Grok 4.5 is the story. #3 overall at 66.7%. Right behind Fable 5 Max at 70.5%. Now look at the cost column. Fable 5 Max: $17.32 per task Grok 4.5 High: $1.51 per task That is Fable level performance at roughly 1/10th the cost. And it https://t.co/W3uwHkQmZr
https://pbs.twimg.com/media/HMuT4OXWsAApBAo.jpg查看原文Dan McAteer ·
Grok 4.5 is a "Wow!" moment. 🤯 > Opus 4.8 & GPT-5.5 level on benchmarks > Costs less than Sonnet 5 Is that real? Hard to believe. Added Grok 4.5 as an implementer agent in fable-advisor. It replaces Sonnet 5. Use Fable as the orchestrator and a combo of: 1. Grok 4.5 2. Opus https://t.co/UOYHjPjXoN
https://pbs.twimg.com/amplify_video_thumb/2074945510662713344/img/tWxeFARi9sH85Sfm.jpg查看原文Reddit
Grok 4.5 is here, and it is clearly aimed at coding agents
Such-Run-4412 ·
展開全文
Grok 4.5 launched as SpaceXAI’s new model for coding, agentic tasks, and knowledge work. It was trained alongside Cursor, which makes the positioning very clear: this model is not just for chat. It is aimed at developers and work automation. The model was trained on coding, science, engineering, and math data, with extra focus on real software engineering tasks. It also used reinforcement learning across hundreds of thousands of tasks, including long agentic rollouts. In simple terms, that means the model was trained to handle longer, multi-step work instead of only answering short prompts. The benchmark numbers are strong, but not a clean sweep. Grok 4.5 scored 62.0% on DeepSWE 1.0, 53% on DeepSWE 1.1, 83.3% on Terminal Bench 2.1, and 64.7% on SWE-Bench Pro. It performs near the top on some coding and terminal tasks, though Fable and GPT-5.5 still lead in several listed comparisons. The speed and pricing are also important. Grok 4.5 runs at 80 tokens per second and is priced at $2 per million input tokens and $6 per million output tokens. It also uses fewer output tokens on SWE-Bench Pro than Opus 4.8, which could make it cheaper for long coding tasks. It is available now in Grok Build, Cursor on all plans, and through the SpaceXAI API console. EU availability is expected in mid-July. Source: https://x.ai/news/grok-4-5
Questions on best way to use Grok 4.5
crzyhorse ·
展開全文
I've been using Grok 4.5 inside of Grok.Build on the free plan in order to try it out. I've been pretty impressed with the speed and the adherence to the idea of what I want. I've been seeing a ton of folks complain about rate limits lately on SuperGrok. I get it, it always sucks but I'm wondering for pure coding, how are the limits? Would I be better off signing up for Cursors $20 plan. Right now I'm only really interested in which one gives me the better limits for Grok 4.5 for coding and especially using Grok.Build.
Grok 4.5 is too fast!
zeleleinspace ·
查看原文YouTube
【2026最新】Grok 4.5を実機検証|料金・無料の使い方とOpus 4.8との差を徹底解説
ITとAIのルーティンラボ
查看原文Grok 4.5 is a bigger deal than Fable 5
Greg Isenberg
查看原文Coding with Grok 4.5 is surprisingly good…
ForrestKnight
查看原文