Side-by-side comparison · models
Grok 4.20 vs Grok 4.3
What we have checked about each, line by line. Every value shows where it came from and when we last confirmed it.
Models
Grok 4.20
Grok 4.20 is a reasoning model from xAI with industry-leading speed and agentic tool calling capabilities.
Models
Grok 4.3
Grok 4.3 is a reasoning model from xAI.
Checked facts
Line by line
| Measure | Grok 4.20 | Grok 4.3 |
|---|---|---|
| Input price | $1.25 per million tokens | $1.25 per million tokens |
| Output price | $2.50 per million tokens | $2.50 per million tokens |
| Price on xai | $0.200 per million cached input tokens | $0.200 per million cached input tokens |
| Price on xai | $1.25 per million input tokens | $1.25 per million input tokens |
| Price on xai | $2.50 per million output tokens | $2.50 per million output tokens |
| Latest version | x-ai/grok-4.20-20260309 | x-ai/grok-4.3-20260430 |
| Time to first token (ms) on xai | 651 | 1,206 |
| Output speed (tokens a second) on xai | 72 | 74 |
| Time to first token (ms) | Not checked yet | 18,390 |
| Output speed (tokens a second) | Not checked yet | 81 |
| Tokens served on OpenRouter in a day | 3.1B (for 10 Nov 2026) | 6.5B (for 10 Nov 2026) |
| Tool calls on OpenRouter in a day | 40.2K (for 10 Nov 2026) | 88.5K (for 10 Nov 2026) |
| Artificial Analysis Intelligence Index | 25.7 | 24.9 |
| Design Arena rating (Elo) | 1,226 | 1,192 |
| Gpqa diamond | 89.3% (for 17 Feb 2026) | 88.8% (for 17 Apr 2026) |
| Frontiermath tiers 1 3 v2 private | 44.9% (for 17 Feb 2026) | 42.8% (for 17 Apr 2026) |
| Terminal bench | 57.3% (for 17 Feb 2026) | Not checked yet |
| Simpleqa verified | 30.2% (for 17 Feb 2026) | 33.2% (for 17 Apr 2026) |
| Arc agi 2 | 65.1% (for 17 Feb 2026) | Not checked yet |
| Arc agi | 89.5% (for 17 Feb 2026) | Not checked yet |
| Chess puzzles | 29% (for 17 Feb 2026) | 18% (for 17 Apr 2026) |
| Cl bench | 22.2% (for 17 Feb 2026) | Not checked yet |
| Cl bench life | 11.9% (for 17 Feb 2026) | Not checked yet |
| Dtbench | 90.1% (for 17 Feb 2026) | 90.7% (for 17 Apr 2026) |
| Frontiermath tier 4 v2 private | 17.1% (for 17 Feb 2026) | 14.6% (for 17 Apr 2026) |
| Lmca | 38.7% (for 17 Feb 2026) | 38.3% (for 17 Apr 2026) |
| Proofbench | 14% (for 17 Feb 2026) | 11% (for 17 Apr 2026) |
| Weirdml | 52.3% (for 17 Feb 2026) | 49.9% (for 17 Apr 2026) |
| Otis mock aime 2024 2025 | 92.2% (saturated benchmark: the top models all score near the ceiling) (for 17 Feb 2026) | 93.3% (saturated benchmark: the top models all score near the ceiling) (for 17 Apr 2026) |
| Artificial Analysis Coding Index | Not checked yet | 42.2 |
| Artificial Analysis Agentic Index | Not checked yet | 15.5 |
| LMArena text rating (Elo) | Not checked yet | 1,397 |
| LMArena web development rating (Elo) | Not checked yet | 1,357 |
| LMArena vision rating (Elo) | Not checked yet | 1,229 |
A line appears only when at least one record has a value we confirmed from a source. Values we have not confirmed are left out, never guessed. Records of different types share fewer lines.
Add a third
Compare with one more record
How this works
Share it, or build your own
This page works for any two or three records in the directory, of any type. The link stays the same whichever order you name them in, so you can share it as it is. Change the records in the address, for example /directories/vs/claude-code-vs-cursor, or start from our editorial comparisons.