Grok-3 vs Grok 4.5: Specs & Benchmark Comparison

Grok-3 and Grok 4.5 both come from xAI, Grok-3 was released in February 2025, and Grok 4.5 followed 17 months later in July 2026.

We have 5 benchmark results for Grok-3 and 15 benchmark results for Grok 4.5, but they overlap on a single test: GPQA, where Grok 4.5 scores 93.0% against 85.0%.

Grok 4.5 is the cheaper API at $2 per million input tokens and $6 per million output tokens, about 50% below Grok-3 at $3 and $15. Grok 4.5 takes the larger context window at 500K tokens, compared with 128K for Grok-3. Both accept text and images as input.

CharacteristicGrok-3Grok 4.5
CompanyxAIxAI
Release DateFebruary 16, 2025July 16, 2026
Parameters
MultimodalYesYes
Context (input)128K500K
Context (output)8K
Input Price / 1M$3.00$2.00
Output Price / 1M$15.00$6.00
Average Score85.7%52.9%
Benchmarks
GPQA85.0%93.0%

Visual Benchmark Comparison

Grok-3
Grok 4.5
GPQA0.8 vs 0.9
0.8
0.9

Verdict

Grok 4.5 leads in 3 out of 4 comparison categories.

Overall Performance

Both models show comparable average scores: Grok-3 — 0.9, Grok 4.5 — 0.5.

API Cost

Grok 4.5 is 2.3x cheaper: input $2.00/1M vs $3.00/1M tokens.

Context Window

Grok 4.5 supports a larger context: 500K vs 128K tokens.

Recency

Grok 4.5 is newer: released 7/16/2026 vs 2/16/2025.

More About These Models

Related Comparisons

Frequently Asked Questions

Which is better for coding — Grok-3 or Grok 4.5?
Direct comparison on the SWE-Bench benchmark is not available. We recommend reviewing other metrics on the comparison page.
Which model is cheaper — Grok-3 or Grok 4.5?
Grok 4.5 is cheaper for input: $2.00 per 1M tokens vs $3.00.
Which has a larger context window — Grok-3 or Grok 4.5?
Grok 4.5 supports a larger context: 500,000 tokens vs 128,000.

The Grok-3 and Grok 4.5 comparison is updated for 2026. Data includes benchmark results, API pricing, context window size and other specifications. For more detailed information, visit the Grok-3 or Grok 4.5 page. See also the complete list of AI model comparisons.