Model comparison
GPT-5.2 Codex vs Qwen3-4B
GPT-5.2 Codex is the stronger model overall, scoring 42.6 to 31.9 on the Noometry Index.
Last verified . 0 shared benchmarks.
Alibaba (Qwen)Summary
- The widest gap is in agentic & tool use, where GPT-5.2 Codex leads 41.0 to 27.6.
- Qwen3-4B has downloadable open weights; the other is API-only.
Side by side
| GPT-5.2 Codex | Qwen3-4B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 42.6 | 31.9 |
| Released | 2025-12-18 | 2025-04-29 |
| Weights | Proprietary | Open |
| Context window | 400K | — |
| Max output | 128K | — |
| Input $ / M tokens | $1.75 | — |
| Output $ / M tokens | $14 | — |
| Results tracked | 5 | 6 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Not comparable
GPT-5.2 Codex: 45.5 (#71), Qwen3-4B: —
| Benchmark | GPT-5.2 Codex | Qwen3-4B |
|---|---|---|
| SWE-bench Verified (bash only) | 72.8% | — |
| LMArena WebDev | 1339 | — |
| SWE-bench Multilingual | 66.3% | — |
| ALE-Bench | 1,300 | — |
Agentic & Tool Use GPT-5.2 Codex leads
GPT-5.2 Codex: 41.0 (#22), Qwen3-4B: 27.6 (#100)
| Benchmark | GPT-5.2 Codex | Qwen3-4B |
|---|---|---|
| Terminal-Bench | 66.5% | — |
| Berkeley Function Calling Leaderboard | — | 35.7% |
Reasoning Not comparable
GPT-5.2 Codex: —, Qwen3-4B: 19.2 (#268)
| Benchmark | GPT-5.2 Codex | Qwen3-4B |
|---|---|---|
| Chess Puzzles | — | 4% |
Math Not comparable
GPT-5.2 Codex: —, Qwen3-4B: 29.7 (#240)
| Benchmark | GPT-5.2 Codex | Qwen3-4B |
|---|---|---|
| MathArena Final-Answer Competitions | — | 38.5% |
| OTIS Mock AIME 2024-2025 | — | 52.2% |
Knowledge Not comparable
GPT-5.2 Codex: —, Qwen3-4B: 33.0 (#208)
| Benchmark | GPT-5.2 Codex | Qwen3-4B |
|---|---|---|
| GPQA Diamond | — | 52.3% |
| Vectara Hallucination Rate | — | 5.7% |
Frequently asked questions
Is GPT-5.2 Codex better than Qwen3-4B?
GPT-5.2 Codex is the stronger model overall, scoring 42.6 to 31.9 on the Noometry Index.
How many benchmarks do GPT-5.2 Codex and Qwen3-4B share?
0 benchmarks have published results for both models. GPT-5.2 Codex has 5 scored results on Noometry and Qwen3-4B has 6.