#66 xAI: Grok 4.20
medium- Cost
- $0.041
- Time
- 110.3s
- Tokens
- 16,336 tok
X AI
Release: 2026-03-31
Tested on: 2026-05-08 15:32
x-ai/grok-4.20::medium
~1T total (~100B active)MoEClosedEstimated
(medium)
(none)
Summary
Grok 4.20 scores 6.9 on AI BENCHY and ranks #66. It has 10.0 reliability, a 63.2% pass rate, $0.756 total cost, and 14.53s average response time.
Identity note
Grok 4.20 Beta was the preview version of Grok 4.20.
Model facts
Researched on 2026-08-12
- Parameters
- ~1T total (~100B active)
- Architecture
- MoE
- Availability
- Closed
- License
- -
Best estimate from public evidence; the vendor did not disclose every value. xAI does not disclose the architecture or count.
6.9
Consistency
8.3
10.0
$0.756
Total Output Tokens
130,017
Total Input Tokens
0
Input Price
$1.250 / 1M
Output Price
$2.500 / 1M
Flaky tests
4
Flaky tests had mixed outcomes across runs (at least one pass and one fail).
Hamster playing table tennis
Prompt: Create a detailed SVG illustration of a hamster playing table tennis.
Run history
| Tested on | Score | Reliability | Tests Correct | Total Cost | Compare |
|---|---|---|---|---|---|
| 2026-08-14 02:57 Re-test | 7.0 | 10.0 | $0.805 ↓ | Compare | |
| 2026-07-16 23:07 New test added | 7.1 | 10.0 | $0.777 ↓ | Compare | |
| 2026-06-04 14:17 New test added | 7.1 | 10.0 | $0.609 ↓ | Compare | |
| 2026-05-22 00:32 Suite changed | 6.7 | 10.0 | $0.832 | Compare | |
| 2026-05-08 15:32 Suite changed | 6.9 | 10.0 | $0.756 | Current run | |
| 2026-04-11 01:44 First recorded run | 7.0 | N/A | $0.743 | Compare |
This run used a different benchmark suite. Keep suite changes in mind when reading historical movement.
Run comparison
| Run | Benchmark coverage | Score | Consistency | Reliability | Tests Correct | Flaky tests | Total Output Tokens | Total Input Tokens | Total Cost | Response Time (avg) |
|---|---|---|---|---|---|---|---|---|---|---|
| 2026-05-08 15:32 · Suite changed | 57/57 attempts | 6.9 | 8.3 | 10.0 | 10/19 | 4 | 130,017 | 0 | $0.756 | 14.53s |
| 2026-08-14 02:57 · Re-test | 66/66 attempts | 7.0 | 8.2 | 10.0 | 11/22 | 5 | 270,259 | 102,986 | $0.805 | 32.56s |
| Difference | — | -0.1 | +0.1 | 0.0 | -1 | -1 | -140242 | -102986 | -$0.049 | -18033ms |
Benchmark coverage differs: 57/57 attempts (Target: 3 repeats per test) versus 66/66 attempts (Target: 3 repeats per test). Totals and repeat-sensitive metrics are not directly comparable.
These two runs used different benchmark suites, so the deltas reflect both model changes and suite changes.
Charts
Choose the first model, then click a second model to open a side-by-side page.
Top Models by Score
Score vs Total Cost
Response Time (avg)
Score vs Response Time (avg)
Total Output Tokens
Score vs Total Output Tokens
Quick Compare
Grok 4.20mediumvsMercury 2mediumGrok 4.20mediumvsSeed-2.0-MinimediumGrok 4.20mediumvsGPT-5.5noneGrok 4.20mediumvsKimi K2.5mediumGrok 4.20mediumvsGLM 5.1mediumGrok 4.20mediumvsKAT-Coder-Pro V2.5mediumGrok 4.20mediumvsDeepSeek V3.2mediumGrok 4.20mediumvsQwen3.7 FlashmediumGrok 4.20mediumvsGPT-5.6 SolnoneGrok 4.20mediumvsSeed-2.0-Codehigh
Category Breakdown
| Category | Score | Consistency | Tests Correct |
|---|---|---|---|
| Anti-AI Tricks | 8.2 | 7.9 | |
| Coding | 4.3 | 1.1 | |
| Combined | 10.0 | 10.0 | |
| Data parsing and extraction | 10.0 | 10.0 | |
| Domain specific | 5.3 | 10.0 | |
| General Intelligence | 3.9 | 2.6 | |
| Instructions following | 7.3 | 6.0 | |
| Puzzle Solving | 7.7 | 10.0 | |
| Tool Calling | 3.0 | 10.0 | |
| Trivia | 3.0 | 10.0 |