LLM·Dex

Grok 3 vs Grok 4

A complete head-to-head: pricing, context window, benchmarks, modality coverage, and openness, with a programmatic verdict synthesized from the underlying data.

Updated

Grok 3 specs · Grok 4 specs
Verdict by category
  • PriceTie

    Both land near $15.00 / 1M output tokens, call it a wash on cost.

  • Context windowGrok 4

    Grok 4 accepts 256K tokens vs 128K, 2.0× the room for long documents and codebases.

  • BenchmarksTie

    No directly comparable public benchmarks are available for both models, check the spec sheets for individual scores.

  • ModalitiesTie

    Both handle text, vision.

  • OpennessTie

    Both are closed-weight, API-only.

On balance Grok 4 edges ahead, winning 1 of 5 categories against Grok 3's 0. Both land near $15.00 / 1M output tokens, call it a wash on cost. Grok 4 accepts 256K tokens vs 128K, 2.0× the room for long documents and codebases.

No directly comparable public benchmarks are available for both models, check the spec sheets for individual scores. Both target the same set of modalities (text, vision), so the deciding factors are price, context, and raw quality. Both are closed-weight, API-only.

Grok 4 is the newer of the two, released 5 months after Grok 3, which usually means a more recent knowledge cutoff and updated safety post-training. Grok 3 is usually picked for chatbots and research agent workloads, while Grok 4 sees more deployments in reasoning and research agent. If pricing matters more than every last benchmark point, run the numbers in the calculator below before committing.

Side-by-side specs

SpecGrok 3Grok 4
ProviderxAIxAI
ReleasedFeb 2025Jul 2025
Modalitiestext, visiontext, vision
Context window128K tokens256K tokens
Max output,,
Input · 1M$3.00 / 1M tokens$3.00 / 1M tokens
Output · 1M$15.00 / 1M tokens$15.00 / 1M tokens
Knowledge cutoff2024-082024-11
Open weightsNoNo
API availableYesYes

Pricing at scale

What you'd actually pay at typical workloads. Numbers come from each model's published per-million-token rates.

  • Light usage, 100k in / 50k out per day$31.50 vs $31.50
  • Heavy usage, 1M in / 500k out per day$315 vs $315
  • RAG workload, 5M in / 200k out per day$540 vs $540

Light usage, 100k in / 50k out per day: both models cost about $31.50 a month, call it a tie. Heavy usage, 1M in / 500k out per day: both models cost about $315 a month, call it a tie. RAG workload, 5M in / 200k out per day: both models cost about $540 a month, call it a tie.

Price calculator

Estimated spend for the listed models at your usage. Numbers are derived from each model's published per-million-token rates.

  • Grok 3$1.05
  • Grok 4$1.05

Benchmarks compared

Only sourced numbers. Where a benchmark is missing for one model we show the available value rather than fabricating the other.

Grok 3Grok 4
  • GPQA,87.5
Pick Grok 3 if

Grok 3 fits when…

  • Real-time X access
  • Reasoning modes
Pick Grok 4 if

Grok 4 fits when…

  • Real-time X data access
  • Strong on benchmarks at launch
  • Less restrictive content policy than peers
  • Long-context tasks, handles 256K tokens vs 128K for Grok 3.
Don't want either?

Consider Grok 2

xAI's first widely-available model, free on X for Premium subscribers and competitive with the GPT-4 mid-tier of its era.

Frequently asked

  • Is Grok 3 or Grok 4 cheaper?
    Both Grok 3 and Grok 4 are priced equivalently at $15.00 / 1M tokens per million output tokens.
  • Which has the larger context window?
    Grok 4 accepts 256K tokens vs 128K for Grok 3.
  • Is Grok 3 or Grok 4 better for coding?
    Both Grok 3 and Grok 4 are competitive on coding benchmarks. See each model's individual spec page for HumanEval and SWE-bench scores where published. For an opinionated pick, consult our Best LLM for Coding ranking.
  • Are either of these models open source?
    Neither model ships open weights, both are accessible only via their respective providers' APIs.
  • When were Grok 3 and Grok 4 released?
    Grok 3 was released by xAI on 2025-02-17. Grok 4 was released by xAI on 2025-07-09.
Friday digest

The week's AI launches, in your inbox.

One short email every Friday, new models, leaks, and quietly-shipped APIs you missed.