Grok 4.5 vs Grok 4.3: Context, Cost and Which to Use (2026)
Grok 4.5 is xAI's coding and agentic model, added to the API on July 8, 2026, with a 500K token context window and $2.00 input / $6.00 output per million tokens. Grok 4.3 has twice the context at 1M tokens, costs $1.25 / $2.50, adds a no-reasoning mode and supports the Batch API. Both accept text and image input. Neither is xAI's newest model any more: Grok 4.6 (August 12, 2026) built on Grok 4.5, and Grok 4.7 (September 21, 2026) is the model xAI now recommends (docs.x.ai, October 2026). Pick Grok 4.5 for coding and agents that fit in 500K tokens, and Grok 4.3 for long inputs, cheap calls and batch jobs.
This comparison uses xAI's own model pages, pricing page and announcements as of September 15, 2026, linked at the end. It replaces an earlier version of this article from July 2026 that repeated several figures xAI never published, including a parameter count for Grok 4.5 and video input for Grok 4.3. For the full lineup see Grok models explained, and for plan and API prices see Grok pricing and plans.
By Adi Leviim, Co-Founder at Infi Developments. 7+ years building AI productivity tools. Specifications checked against docs.x.ai on September 15, 2026.
What Is the Difference Between Grok 4.5 and Grok 4.3?
Grok 4.5 is built for coding and agentic tasks with a 500K window, while Grok 4.3 is a cheaper general model with a 1M window and more control over reasoning. xAI describes Grok 4.5 as its "intelligent coding model for agentic software, engineering, and workflow tasks," trained on datasets spanning science, engineering and math. It describes Grok 4.3 as a "fast, reliable model with strong tool calling and instruction following capabilities."
The biggest practical differences are the window and the reasoning floor. Grok 4.3 can run with reasoning set to none, which xAI uses as the replacement for its retired non-reasoning models, while Grok 4.5 always reasons at low effort or above.
Grok 4.5 vs Grok 4.3 for Coding and Agents
For coding and agent work, Grok 4.5 is the stronger of the two by xAI's own measurements. xAI says Grok 4.5 was "trained alongside Cursor," with reinforcement learning across hundreds of thousands of multi-step software engineering tasks, and is served at about 80 tokens per second.
Reported scores for Grok 4.5: 83.3% on Terminal Bench 2.1, 64.7% on SWE Bench Pro, 62.0% on DeepSWE 1.0 and 29.0% on SWE Marathon, per xAI's July 16, 2026 announcement.
Token efficiency: xAI reports an average of 15,954 output tokens per SWE Bench Pro task for Grok 4.5, about 4.2 times fewer than Opus 4.8 (max) at 67,020. Fewer output tokens partly offsets its higher per-token price.
Grok 4.3: xAI's migration guide for the May 15, 2026 retirements says Grok 4.3 "tops the leaderboards in agentic tool calling and instruction following"; xAI's announcement for Grok 4.5 does not include Grok 4.3 in its benchmark charts, so there is no official side-by-side score.
All of these figures are published by xAI, not independent tests. For coding today, xAI's models page recommends Grok 4.7 over both. Grok 4.6, the release in between, scores 61 on the Artificial Analysis Intelligence Index against 56 for Grok 4.5 in xAI's evaluation table.
Context Window: Grok 4.3 Holds Twice as Much
Grok 4.3 accepts up to 1M tokens in one request, twice Grok 4.5's 500K. If you send long contracts, research folders or large codebases in a single call, Grok 4.3 (or Grok 4.20, also 1M) fits inputs that Grok 4.5 cannot. xAI does not publish a reason for the smaller window on Grok 4.5.
Filling either window costs more than the headline price: once a prompt reaches 200K tokens, xAI bills every token in the request at the long context rate, which is double the standard rate on both models. The 500K and 1M figures are API limits; xAI does not publish a token figure for conversations in the Grok apps.
In long Grok chats, the AI Toolbox context window meter estimates how full a conversation is so you can start a fresh one before earlier messages drop out. To size a prompt first, use the token counter.
Cost: Which Is Cheaper, Grok 4.5 or Grok 4.3?
Grok 4.3 is cheaper on every token type: $1.25 against $2.00 for input, $0.20 against $0.30 for cached input, and $2.50 against $6.00 for output. Worked from xAI's published rates, a request with 100K input tokens and 5K output tokens costs about $0.23 on Grok 4.5 and about $0.14 on Grok 4.3. Output-heavy work widens the gap, since Grok 4.5's output rate is 2.4 times higher.
Batch jobs: Grok 4.3 gets a 20% Batch API discount; Grok 4.5 does not accept batch requests.
Retired models: since May 15, 2026, calls to retired names such as grok-4-1-fast-reasoning, grok-4-0709 and grok-3 are served by Grok 4.3 and billed at Grok 4.3 rates.
Priority Processing: doubles the rate on either model.
Which One Should You Use?
Choose by task size and budget, and consider Grok 4.7 before either.
Software engineers building coding agents: Grok 4.7 first, since it has the same 500K window and the same input and output prices as Grok 4.5. Grok 4.5 still makes sense for prompts that lean heavily on caching, where its cached input rate is $0.30 against $0.50 on Grok 4.7 and Grok 4.6, or for workflows you have already tested on it.
Legal and research teams sending very long documents: Grok 4.3, for the 1M window.
Developers running classification, extraction or summaries at volume: Grok 4.3 with reasoning set to none or low, plus the Batch API discount.
Anyone migrating off Grok 4.1 Fast or Grok 4: Grok 4.3 is xAI's documented replacement; set the reasoning effort explicitly rather than relying on the redirect.
People using grok.com, X or the Grok apps: you are not choosing between these two. xAI's plan comparison lists Grok 4.6 on every plan, including Free.
Using AI Toolbox With Either Grok Model
AI Toolbox (formerly ChatGPT Toolbox) is a Chrome extension that works inside Grok on x.com/i/grok and grok.com, independent of the model behind it. It adds full-text search across your synced Grok messages, folders and Smart Tags, Context Mentions (@@), a prompt library, the context meter, and export as TXT, Markdown, JSON or PDF. The free plan includes 2 folders, 5 search results per search and TXT export of any single chat; Premium ($9.99/month or $99 lifetime) adds unlimited folders and results, export as Markdown, JSON and PDF, and bulk export and delete. The Grok module purchase is linked to your X handle.
Whichever Grok model you use, keep your chats findable. AI Toolbox adds full-text search, folders and export to Grok on X and grok.com. 40,000+ users, 4.7/5 on the Chrome Web Store. Compare AI Toolbox Grok plans
Limitations and What xAI Documents Well
xAI's model pages make the API comparison easy to verify: context, reasoning settings, prices, batch support and rate limits are all published per model. What xAI does not publish is a release date for Grok 4.3, a parameter count or architecture for either model, or a benchmark chart that includes both. Earlier claims you may see elsewhere, such as a 1.5-trillion-parameter design for Grok 4.5, video input or document generation for Grok 4.3, or a 2M window for Grok 4.20, do not appear on xAI's current model or pricing pages. AI Toolbox does not change which model Grok uses.
Frequently Asked Questions
Is Grok 4.5 better than Grok 4.3?
For coding and agentic tasks, yes: xAI built Grok 4.5 for that work and reports strong SWE Bench Pro and Terminal Bench scores. Grok 4.3 is better for inputs over 500K tokens, lower cost, no-reasoning calls and batch jobs. Both have since been overtaken by Grok 4.6 and then Grok 4.7, which xAI now recommends.
Why does Grok 4.5 have a smaller context window than Grok 4.3?
xAI does not explain it. Its model pages simply list 500K tokens for Grok 4.5 and 1M tokens for Grok 4.3. If your inputs exceed 500K tokens, use Grok 4.3 or Grok 4.20, which also holds 1M. Grok 4.6 and Grok 4.7 keep the 500K window of Grok 4.5.
Is Grok 4.5 more expensive than Grok 4.3?
Yes. Grok 4.5 costs $2.00 per million input tokens, $0.30 cached and $6.00 output, while Grok 4.3 costs $1.25, $0.20 and $2.50. Both double once a prompt reaches 200K tokens. Grok 4.3 also gets a 20% Batch API discount, which Grok 4.5 does not support.
Which Grok model is best for coding?
xAI's models page now recommends Grok 4.7 for code, released September 21, 2026. Between the two older models, Grok 4.5 is the coding model, trained alongside Cursor according to xAI. Grok Build 0.1 is xAI's early-access coding model with a 256K window.
Are Grok 4.5 and Grok 4.3 still available?
Yes, both are listed on xAI's API pricing page in September 2026. Grok 4.3 is also where retired models such as Grok 4.1 Fast and Grok 3 now redirect. In the Grok apps and on X, xAI's plans list Grok 4.6 as the model on every plan.
Does AI Toolbox work with both Grok 4.5 and Grok 4.3?
Yes. AI Toolbox runs in your browser on x.com/i/grok and grok.com, so it does not depend on the model or plan you use. It adds full-text search, folders, Smart Tags, a prompt library, a context meter and export: TXT of any single chat for free, and Markdown, JSON and PDF on Premium.
A Full Stack Developer with 7+ years of experience building AI productivity tools. Leads product development and frontend architecture for AI Toolbox, the Chrome extension suite (ChatGPT, Gemini, Grok, and Claude modules) that helps users search, organize, and export their AI conversations.
Liked this guide? Get the next one first.
Join 40,000+ readers getting practical prompts, new features, and subscriber-only deals. No spam, unsubscribe in one click.
New features and product updates before anyone else
Time-saving prompts and workflows for ChatGPT, Gemini, Claude, and Grok
Adi Leviim