xAI model product
Grok 4
' Built on version 6 of xAI's foundation model, it uses 100x more training compute than Grok 2 and 10x more reinforcement learning compute than Grok 3.
Updated Aug 17, 2026. Default version: Grok-4
LLMBoard Score56.1Grok-4
Coverage80%7 benchmark families
Context window256KTokens
Official input priceN/AOfficial price unavailable
Grok 4 Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Grok 4 Benchmark Results
Benchmark scores for Grok-4.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkUSAMO25 | Score37.5% | Rank03 | Participants3 | Percentile0.0% | EvidenceC | Evaluated |
| BenchmarkHMMT25 | Score90.0% | Rank06 | Participants25 | Percentile79.2% | EvidenceC | Evaluated |
| BenchmarkARC-AGI v2 | Score15.9% | Rank14 | Participants17 | Percentile18.8% | EvidenceC | Evaluated |
| BenchmarkLiveCodeBench | Score79.0% | Rank15 | Participants75 | Percentile81.1% | EvidenceC | Evaluated |
| BenchmarkHumanity's Last Exam | Score40.0% | Rank37 | Participants99 | Percentile63.3% | EvidenceC | Evaluated |
| BenchmarkAIME 2025 | Score91.7% | Rank40 | Participants115 | Percentile65.8% | EvidenceC | Evaluated |
| BenchmarkGPQA | Score87.5% | Rank42 | Participants239 | Percentile82.8% | EvidenceC | Evaluated |
Grok 4 Arena Results
Preference and agent-evaluation results for the default version.
Sort by
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenasearch | Categoryoverall | Rank26 | Rating / score1141.3 | Votes19,379 | ObservationsN/A | Result date |
| Arenasearch factuality | Categoryoverall | Rank28 | Rating / score1131.3 | Votes3,149 | ObservationsN/A | Result date |
| Arenasearch style control | Categoryoverall | Rank30 | Rating / score1121.5 | Votes19,379 | ObservationsN/A | Result date |
| Arenavision | Categoryoverall | Rank69 | Rating / score1208.9 | Votes32,273 | ObservationsN/A | Result date |
| Arenavision style control | Categoryoverall | Rank83 | Rating / score1181.8 | Votes32,273 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryoverall | Rank113 | Rating / score1418.4 | Votes10,658 | ObservationsN/A | Result date |
| Arenatext | Categoryoverall | Rank119 | Rating / score1409.7 | Votes41,551 | ObservationsN/A | Result date |
| Arenatext style control | Categoryoverall | Rank129 | Rating / score1409.2 | Votes41,551 | ObservationsN/A | Result date |
Grok 4 Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- From $3 input, $15 output per 1M via FastRouter
- Tracked offerings
- 6
Sort by
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderFastRouter | Provider model IDx-ai/grok-4 | Regionglobal | Input / 1M$3 | Output / 1M$15 | Context256K | Updated |
| ProviderZenMux | Provider model IDx-ai/grok-4 | Regionglobal | Input / 1M$3 | Output / 1M$15 | Context256K | Updated |
| ProviderHelicone | Provider model IDgrok-4 | Regionglobal | Input / 1M$3 | Output / 1M$15 | Context256K | Updated |
| ProviderPoe | Provider model IDxai/grok-4 | Regionglobal | Input / 1M$3 | Output / 1M$15 | Context256K | Updated |
| ProviderLLM Gateway | Provider model IDgrok-4 | Regionglobal | Input / 1M$3 | Output / 1M$15 | Context256K | Updated |
| ProviderAbacus | Provider model IDgrok-4-0709 | Regionglobal | Input / 1M$3 | Output / 1M$15 | Context256K | Updated |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Grok 4 Runtime Performance
Provider-specific output speed and catalog latency for Grok-4. Runtime does not affect the capability score.
Sort by
Provider | Output Speed | Catalog Latency | Max Input | Max Output | Updated |
|---|
| ProviderxAI | Output Speed100 tok/s | Catalog Latency0.7 s | Max Input256K | Max Output8K | Updated |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Grok 4 Specifications
Technical details for the model's default version.
- Version
- Grok-4
- Released
- Jul 9, 2025
- Knowledge cutoff
- Dec 31, 2024
- Parameters
- N/A
- Context window
- 256K
- Max output
- 8K
- Inputs
- image, text
- Outputs
- text
- Open weights
- No
- License
- Proprietary
Grok 4 Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
Sort by
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionGrok-4 | Released | LLMBoard56.1 | ParametersN/A | Context256K | Max output8K | Open weightsNo | LicenseProprietary |
Grok 4 vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to Grok 4
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
XA
Grok 4.20 Beta Reasoning
xAI
XA
Grok 3
xAI
XA
Grok 4 Fast
xAI
XA
Grok 4.1 Thinking
xAI
XA
Grok 3 Mini
xAI
XA
Grok 4.1
xAI
What is Grok 4?
Key information about Grok 4 and its available data.
' Built on version 6 of xAI's foundation model, it uses 100x more training compute than Grok 2 and 10x more reinforcement learning compute than Grok 3.
The model achieves PhD-level performance across all academic disciplines simultaneously, scoring perfect on standardized tests like the SAT and near-perfect on graduate exams like the GRE. Unlike Grok 3, tool usage is built into the training process rather than relying on generalization.
Trained using 200,000 GPUs, Grok 4 excels at complex reasoning, mathematical problem-solving, and coding tasks, though it has acknowledged weaknesses in multimodal capabilities that are being addressed in the next version.
Data as of 2026-08-17.
FAQ
Common questions about Grok 4.
When was Grok 4 released?
Grok 4's default version was released on Jul 9, 2025.
How much does Grok 4 cost?
No official standard PAYG price is currently available for Grok 4. The lowest tracked third-party offer starts at $3 input and $15 output via FastRouter.
Who created Grok 4?
Grok 4 was created by xAI.
What is the context window for Grok 4?
The default version has a 256K token context window.
Is Grok 4 open weight?
No. The default version is not marked as having publicly available weights.
How many API providers offer Grok 4?
6 provider offerings are linked to the default version.
What models should I compare Grok 4 with?
Nearby ranked alternatives include Qwen3.5 27B, Claude Sonnet 4.5, GLM 5V Turbo.