Side-by-side comparison · models
Qwen2.5 72B Instruct vs Qwen 3 235b A22B 2507 (TEE)
What we have checked about each, line by line. Every value shows where it came from and when we last confirmed it.
Models
Qwen2.5 72B Instruct
Qwen2.5 72B is the latest series of Qwen large language models.
Qwen instruction model for multilingual chat, reasoning, and tool use.
Checked facts
Line by line
| Measure | Qwen2.5 72B Instruct | Qwen 3 235b A22B 2507 (TEE) |
|---|---|---|
| Input price | $0.360 per million tokens | Not checked yet |
| Output price | $0.400 per million tokens | Not checked yet |
| Price on deepinfra | $0.360 per million input tokens | Not checked yet |
| Price on deepinfra | $0.400 per million output tokens | Not checked yet |
| Latest version | qwen/qwen-2.5-72b-instruct | Not checked yet |
| Time to first token (ms) on deepinfra | 1,053 | Not checked yet |
| Output speed (tokens a second) on deepinfra | 21 | Not checked yet |
| Time to first token (ms) | Not checked yet | 840 |
| Output speed (tokens a second) | Not checked yet | 106.6 |
| Tokens served on OpenRouter in a day | 203.7M (for 10 Nov 2026) | Not checked yet |
| Tool calls on OpenRouter in a day | 7.6K (for 10 Nov 2026) | Not checked yet |
| Artificial Analysis Intelligence Index | 7.7 | Not checked yet |
| LMArena text rating (Elo) | 1,269 | Not checked yet |
| Gpqa diamond | 49.1% (for 19 Sept 2024) | Not checked yet |
| Balrog | 16.2% (for 19 Sept 2024) | Not checked yet |
| Dtbench | 62.9% (for 19 Sept 2024) | Not checked yet |
| Lmca | 13.4% (for 19 Sept 2024) | Not checked yet |
| Math level 5 | 63.2% (for 19 Sept 2024) | Not checked yet |
| Metr time horizons | 35.8% (for 19 Sept 2024) | Not checked yet |
| Mmlu | 85.3% (for 19 Sept 2024) | Not checked yet |
| The agent company | 5.7% (for 19 Sept 2024) | Not checked yet |
| Weirdml | 16% (for 19 Sept 2024) | Not checked yet |
| Otis mock aime 2024 2025 | 8.1% (saturated benchmark: the top models all score near the ceiling) (for 19 Sept 2024) | Not checked yet |
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.