Pinned coding agents
Teams that tested Grok 4.5 against their repositories and want a fixed model ID instead of silently changing behavior.
Independent tool overview
Grok 4.5 at a glance
Grok 4.5 is an active SpaceXAI reasoning model for coding, agentic engineering and knowledge work, though Grok 4.6 has replaced it as the newer flagship option.
Visit the official Grok 4.5 site ↗
- Model ID
- grok-4.5
- Current status
- Active; succeeded by Grok 4.6
- Primary focus
- Coding, agents, engineering and knowledge work
- Input
- Text and images
- Output
- Text
- Context window
- 500,000 tokens
- Base API price
- $2 input / $6 output per 1M tokens
- Long-context threshold
- 200,000 prompt tokens
- Batch API
- Not supported
Overview
What Grok 4.5 is
Grok 4.5 remains listed in the SpaceXAI API with a fixed grok-4.5 model ID, a 500,000-token context window, text and image input, text output, configurable reasoning, function calling and structured outputs. It was built for agentic software engineering and technical knowledge work and is also available through products and partners including Grok Build, Cursor and GitHub Copilot.
It is no longer the newest Grok model. SpaceXAI released Grok 4.6 on August 12, 2026 with stronger long-running agent and visual-work capabilities at the same base $2 input and $6 output rates. Grok 4.5 still makes sense when a team has already validated its behavior or needs a pinned version, but new evaluations should compare 4.6 before standardizing.
Use cases
Who Grok 4.5 is best for
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
Large-codebase analysis
Engineering workflows using a 500K context window to inspect broad code and documentation sets, while watching the 200K pricing threshold.
Tool-using technical agents
Applications combining reasoning with function calls, structured output, web or X search, code execution and private collections.
Multimodal debugging
Prompts that pair source code or logs with screenshots, diagrams or other image input and require a text response.
Model migration baselines
Teams comparing an established Grok 4.5 evaluation set against Grok 4.6 before switching production traffic.
Capabilities
Core Grok 4.5 features
Agentic coding model
Designed for multi-step software, engineering and workflow tasks rather than only short conversational answers.
500K context window
Handles large prompts and histories, with higher token rates applied to the entire request once the prompt reaches 200,000 tokens.
Image understanding
Accepts images alongside text for tasks such as reviewing screenshots, diagrams and visual technical context.
Configurable reasoning
Supports low, medium or high reasoning effort so applications can trade latency and token use against deeper analysis.
Function calling
Can select and call application-defined functions to retrieve data or take actions under developer-controlled permissions.
Structured outputs
Can produce responses constrained to a defined schema for more reliable downstream application handling.
Server-side tools
The Responses API can add web search, X search, code execution, file attachments, collections search, image generation and remote MCP tools at separate rates.
Prompt caching
Cached input tokens cost less than regular input when repeated prompt prefixes qualify for cache reuse.
Process
How the Grok 4.5 workflow works
Step 1
Choose 4.5 deliberately
Use the fixed grok-4.5 ID when preserving evaluated behavior matters; benchmark Grok 4.6 first for a new implementation.
Step 2
Bound the context
Retrieve only the code, documentation and history needed for the task, because crossing 200K prompt tokens doubles all Grok 4.5 token rates.
Step 3
Define tools and schemas
Expose narrow functions, validate arguments, require structured output where useful and keep authorization outside the model.
Step 4
Set reasoning effort
Use lower effort for routine work and reserve higher effort for complex debugging, architecture or multi-step execution.
Step 5
Run in an isolated environment
Give coding agents scoped credentials, a disposable branch or sandbox and explicit limits on commands, networks and secrets.
Step 6
Measure the real request cost
Track input, cached, reasoning and output tokens plus every server-side tool call; do not estimate an agent run from headline token rates alone.
Step 7
Verify before merging or acting
Require tests, static checks, diff review and human approval for consequential code, infrastructure, financial or business actions.
Cost
Grok 4.5 pricing and free plan
Grok 4.5 uses token-based API pricing. Prompts below 200K tokens cost $2 per million input tokens, $0.30 per million cached input tokens and $6 per million output tokens. Once the prompt reaches 200K tokens, those rates double for all tokens in the request. Server-side tools and optional priority processing add separate charges.
Short-context API
$2 input / $0.30 cached / $6 output per 1M tokens
Standard Grok 4.5 pricing for prompts below 200,000 tokens.
- 500,000-token maximum context window
- Reasoning tokens are billed as output
- Fixed model ID is grok-4.5
- Batch API is not supported
Long-context API
$4 input / $0.60 cached / $12 output per 1M tokens
Rates applied to every token in a request when the prompt reaches 200,000 tokens.
- The higher rate is not limited to tokens above the threshold
- Context retrieval and compaction can materially reduce cost
- Monitor the assembled prompt rather than only the user's latest message
Server-side tools
From $2.50 to $10 per 1,000 calls, plus tokens
Optional tools are billed per invocation in addition to model token use.
- Web Search, X Search and code execution: $5 per 1,000 calls
- Collections search: $2.50 per 1,000 calls
- File attachment search: $10 per 1,000 calls
- Priority processing is 2x standard token pricing when actually served at priority
Pricing checked . Check current pricing at the source ↗
Assessment
Grok 4.5 strengths and limitations
Where it stands out
- The 500K context window supports broad repository and knowledge-work prompts.
- Text and image input cover code, documentation, screenshots and diagrams in one request.
- Function calling and structured outputs make it suitable for application and agent workflows.
- Configurable reasoning lets developers tune routine and difficult requests differently.
- Prompt caching can reduce repeated-context cost to $0.30 per million tokens below the long-context threshold.
- A fixed model ID gives teams a stable migration baseline even after Grok 4.6's release.
What to consider
- Grok 4.6 is now the newer flagship and is positioned for stronger long-running agents and more ambitious visual or interactive work at the same base rates.
- Crossing 200,000 prompt tokens doubles the input, cached-input and output rates for the entire request.
- Grok 4.5 has text output only; image input does not mean it generates images through the base model response.
- The Batch API is not supported for Grok 4.5.
- Server-side search, code execution and file tools create charges beyond model tokens, and autonomous agents may invoke them multiple times.
- Reasoning and long agent loops can make output-token costs larger than a simple chat estimate suggests.
- Model-generated code can introduce security, dependency, licensing and logic defects even when it passes a superficial review.
- Function calling does not provide authorization; the application must validate arguments and restrict what every tool can do.
- Vendor benchmark results do not predict performance on a team's own repositories, languages, frameworks or agent harness.
- Aliases can change routing over time, so production systems that need reproducibility should use a fixed model ID and monitor deprecation notices.
Compare
Grok 4.5 alternatives
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Consumer
Grok 4.6
Choose Grok 4.6 for SpaceXAI's newer flagship, especially for longer-running agents and interactive or visual application work.
Explore Grok 4.6 →Coding
Grok Build
Choose Grok Build when you want SpaceXAI's ready-made coding agent and workflow environment instead of integrating the model API yourself.
Explore Grok Build →Coding
Claude Code
Choose Claude Code for an established terminal and IDE coding-agent workflow with Anthropic models.
Explore Claude Code →Coding
Cursor
Choose Cursor for an editor-first coding product that can expose Grok and competing models inside one development workflow.
Explore Cursor →Questions
Grok 4.5 FAQs
Is Grok 4.5 still available?
Yes. SpaceXAI still lists grok-4.5 as an active API model with its own documentation and pricing. It is no longer the newest flagship.
Has Grok 4.5 been replaced by Grok 4.6?
Grok 4.6 is the newer flagship and builds on 4.5, but Grok 4.5 remains available. Existing teams can keep the fixed model while evaluating migration.
How much does the Grok 4.5 API cost?
Below 200K prompt tokens, the rates are $2 per million input tokens, $0.30 per million cached tokens and $6 per million output tokens. Long-context rates are double.
What happens above 200K tokens?
When the prompt reaches 200,000 tokens, SpaceXAI charges $4 input, $0.60 cached input and $12 output per million tokens for all tokens in that request.
What is Grok 4.5's context window?
The documented context window is 500,000 tokens. The 200K pricing threshold is lower than the maximum context size.
Can Grok 4.5 analyze images?
Yes. The model accepts text and image input and returns text. Image generation is a separate server-side tool and Imagine API product.
Does Grok 4.5 support function calling?
Yes. It supports function calling and structured outputs, but developers must validate requests, enforce permissions and confirm consequential actions outside the model.
Does Grok 4.5 support batch processing?
No. Its model page lists the Batch API as unsupported.
Are web and X search included in the token price?
No. Web Search and X Search each cost $5 per 1,000 tool calls in addition to the model tokens used by the request.
Should a new project use Grok 4.5 or 4.6?
Start by evaluating Grok 4.6 because it is newer and has the same base token rates. Keep 4.5 when its behavior is already validated or it performs better on a specific workload.
Bottom line
Our Grok 4.5 verdict
Grok 4.5 remains a capable, unusually fast API option for coding and tool-using technical agents, with a large context window and straightforward base rates. Its value now is a stable, pinned model rather than the newest Grok experience. New deployments should benchmark Grok 4.6, and every deployment should control the 200K long-context threshold, tool-call amplification and permissions around model-selected actions.
Visit Grok 4.5 website ↗Similar Consumer Tools
View all Consumer AI tools →
Willow Frontier Mini
Willow Frontier Mini - Willow's free, unlimited AI dictation model
GPT-Live
GPT-Live - OpenAI's new voice model with more natural conversation abilities
Hy-3
Hy-3 - Tencent's new open-source model built for cheap, reliable agent deployment
Muse Spark 1.1
Muse Spark 1.1 - Meta's upgraded, cost-effective agentic model with 1M-token memory and computer use
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.