Model comparison
GPT-5.3 Chat vs Hy4 preview
Hy4 preview is the stronger model overall, scoring 45.3 to 42.8 on the Noometry Index.
Last verified . 0 shared benchmarks.
TencentSummary
- The widest gap is in math, where Hy4 preview leads 55.7 to 38.2.
- Hy4 preview is cheaper at $0.83 / $2.50 per million input/output tokens, against $1.75 / $14 for GPT-5.3 Chat.
- Hy4 preview accepts more context: 1.05M tokens versus 128K.
- Hy4 preview has downloadable open weights; the other is API-only.
Side by side
| GPT-5.3 Chat | Hy4 preview | |
|---|---|---|
| Provider | OpenAI | Tencent |
| Noometry Index | 42.8 | 45.3 |
| Released | 2026-03-03 | 2026-08-28 |
| Weights | Proprietary | Open |
| Context window | 128K | 1.05M |
| Max output | 16K | 64K |
| Input $ / M tokens | $1.75 | $0.83 |
| Output $ / M tokens | $14 | $2.50 |
| Results tracked | 18 | 3 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Hy4 preview leads
GPT-5.3 Chat: 41.4 (#124), Hy4 preview: 51.6 (#38)
| Benchmark | GPT-5.3 Chat | Hy4 preview |
|---|---|---|
| LMArena WebDev | — | 1632 |
| LMArena Coding | 1408 | — |
Reasoning Hy4 preview leads
GPT-5.3 Chat: 28.5 (#102), Hy4 preview: 31.9 (#79)
| Benchmark | GPT-5.3 Chat | Hy4 preview |
|---|---|---|
| NYT Connections (extended) | — | 68.2% |
| LMArena Hard Prompts | 1399 | — |
Math Hy4 preview leads
GPT-5.3 Chat: 38.2 (#142), Hy4 preview: 55.7 (#42)
| Benchmark | GPT-5.3 Chat | Hy4 preview |
|---|---|---|
| ProofBench | — | 75% |
| LMArena Math | 1389 | — |
Knowledge Not comparable
GPT-5.3 Chat: 38.8 (#140), Hy4 preview: —
| Benchmark | GPT-5.3 Chat | Hy4 preview |
|---|---|---|
| LMArena Expert | 1397 | — |
Multilingual Not comparable
GPT-5.3 Chat: 50.3 (#124), Hy4 preview: —
| Benchmark | GPT-5.3 Chat | Hy4 preview |
|---|---|---|
| LMArena Non-English | 1382 | — |
| LMArena Chinese | 1432 | — |
| LMArena French | 1397 | — |
| LMArena German | 1384 | — |
| LMArena Japanese | 1352 | — |
| LMArena Korean | 1346 | — |
| LMArena Russian | 1400 | — |
| LMArena Spanish | 1371 | — |
Instruction Following Not comparable
GPT-5.3 Chat: 72.8 (#129), Hy4 preview: —
| Benchmark | GPT-5.3 Chat | Hy4 preview |
|---|---|---|
| LMArena Instruction Following | 1378 | — |
Long Context Not comparable
GPT-5.3 Chat: 42.6 (#120), Hy4 preview: —
| Benchmark | GPT-5.3 Chat | Hy4 preview |
|---|---|---|
| LMArena Longer Query | 1396 | — |
Writing & Preference Not comparable
GPT-5.3 Chat: 63.1 (#68), Hy4 preview: —
| Benchmark | GPT-5.3 Chat | Hy4 preview |
|---|---|---|
| LMArena Text | 1389 | — |
| LMArena Creative Writing | 1355 | — |
| EQ-Bench Creative Writing | 1690 | — |
| LMArena Multi-Turn | 1412 | — |
Frequently asked questions
Is GPT-5.3 Chat better than Hy4 preview?
Hy4 preview is the stronger model overall, scoring 45.3 to 42.8 on the Noometry Index.
Which is cheaper, GPT-5.3 Chat or Hy4 preview?
Hy4 preview is cheaper. It lists at $0.83 per million input tokens and $2.50 per million output tokens; GPT-5.3 Chat lists at $1.75 and $14.
Is GPT-5.3 Chat or Hy4 preview better for coding?
Hy4 preview scores higher on coding benchmarks: 51.6 versus 41.4 in the Noometry coding category.
Which has the bigger context window?
Hy4 preview does, with 1.05M tokens against 128K.
How many benchmarks do GPT-5.3 Chat and Hy4 preview share?
0 benchmarks have published results for both models. GPT-5.3 Chat has 18 scored results on Noometry and Hy4 preview has 3.