Side-by-side comparison · models
GPT-4.1 vs GPT-4.1 Mini
What we have checked about each, line by line. Every value shows where it came from and when we last confirmed it.
Models
GPT-4.1
GPT-4.1 is a flagship large language model optimized for advanced instruction following, real-world software engineering, and long-context reasoning.
Models
GPT-4.1 Mini
GPT-4.1 Mini is a mid-sized model delivering performance competitive with GPT-4o at substantially lower latency and cost. It retains a 1 million token context window and scores 45.
Checked facts
Line by line
| Measure | GPT-4.1 | GPT-4.1 Mini |
|---|---|---|
| Input price | $2.00 per million tokens | $0.400 per million tokens |
| Output price | $8.00 per million tokens | $1.60 per million tokens |
| Price on azure | $0.500 per million cached input tokens | $0.100 per million cached input tokens |
| Price on azure | $2.00 per million input tokens | $0.400 per million input tokens |
| Price on azure | $8.00 per million output tokens | $1.60 per million output tokens |
| Price on openai | $0.500 per million cached input tokens | $0.100 per million cached input tokens |
| Price on openai | $2.00 per million input tokens | $0.400 per million input tokens |
| Price on openai | $8.00 per million output tokens | $1.60 per million output tokens |
| Latest version | openai/gpt-4.1-2025-04-14 | openai/gpt-4.1-mini-2025-04-14 |
| Time to first token (ms) | 1,880 | 1,620 |
| Time to first token (ms) on azure | 1,113 | 2,623 |
| Time to first token (ms) on openai | 950 | 878 |
| Output speed (tokens a second) | 161.5 | 156.1 |
| Output speed (tokens a second) on azure | 59 | 75 |
| Output speed (tokens a second) on openai | 65 | 81 |
| Tokens served on OpenRouter in a day | 8.7B (for 10 Nov 2026) | 18B (for 10 Nov 2026) |
| Tool calls on OpenRouter in a day | 222.7K (for 10 Nov 2026) | 235.6K (for 10 Nov 2026) |
| Artificial Analysis Intelligence Index | 12.7 | 10.2 |
| LMArena text rating (Elo) | 1,383 | 1,341 |
| LMArena vision rating (Elo) | 1,210 | 1,181 |
| Design Arena rating (Elo) | 1,104 | 1,064 |
| Gpqa diamond | 66.9% (for 14 Apr 2025) | 65.8% (for 14 Apr 2025) |
| Swe bench verified | 48.5% (for 14 Apr 2025) | Not checked yet |
| Frontiermath tiers 1 3 v2 private | 6% (for 14 Apr 2025) | 6.7% (for 14 Apr 2025) |
| Simpleqa verified | 31.1% (for 14 Apr 2025) | 12.7% (for 14 Apr 2025) |
| Arc agi 2 | 0.4% (for 14 Apr 2025) | 0% (for 14 Apr 2025) |
| Aider polyglot | 52.4% (for 14 Apr 2025) | 32.4% (for 14 Apr 2025) |
| Arc agi | 5.5% (for 14 Apr 2025) | 3.5% (for 14 Apr 2025) |
| Cadeval | 42% (for 14 Apr 2025) | 16% (for 14 Apr 2025) |
| Chess puzzles | 3% (for 14 Apr 2025) | 1% (for 14 Apr 2025) |
| Dtbench | 68.3% (for 14 Apr 2025) | 68.8% (for 14 Apr 2025) |
| Fiction livebench | 63.9% (for 14 Apr 2025) | 44.4% (for 14 Apr 2025) |
| Frontiermath 2025 02 28 private | 5.5% (for 14 Apr 2025) | 4.5% (for 14 Apr 2025) |
| Frontiermath tier 4 2025 07 01 private | 0% (for 14 Apr 2025) | Not checked yet |
| Geobench | 72% (for 14 Apr 2025) | Not checked yet |
| Lmca | 25.6% (for 14 Apr 2025) | 21.1% (for 14 Apr 2025) |
| Math level 5 | 83% (for 14 Apr 2025) | 87.3% (for 14 Apr 2025) |
| Simplebench | 27% (for 14 Apr 2025) | Not checked yet |
| Weirdml | 39% (for 14 Apr 2025) | 37.6% (for 14 Apr 2025) |
| Otis mock aime 2024 2025 | 38.3% (saturated benchmark: the top models all score near the ceiling) (for 14 Apr 2025) | 44.7% (saturated benchmark: the top models all score near the ceiling) (for 14 Apr 2025) |
| Artificial Analysis Coding Index | Not checked yet | 20.2 |
| Mystery game puzzles | Not checked yet | 7% (for 14 Apr 2025) |
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.