Model comparison
GPT-6 Astra vs Grok 4.3
GPT-6 Astra is the stronger model overall, scoring 70.8 to 43.8 on the Noometry Index. Grok 4.3 costs 13× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Last verified . 37 shared benchmarks.
Summary
- They share 37 benchmarks with published results for both. GPT-6 Astra scores higher in 10 categories and Grok 4.3 in 0 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-6 Astra leads 85.1 to 35.9.
- The biggest single-benchmark swing is ProofBench: 99% for GPT-6 Astra and 11% for Grok 4.3.
- Grok 4.3 is cheaper at $1.25 / $2.50 per million input/output tokens, against $10 / $50 for GPT-6 Astra.
- GPT-6 Astra accepts more context: 1.05M tokens versus 1M.
Side by side
| GPT-6 Astra | Grok 4.3 | |
|---|---|---|
| Provider | OpenAI | xAI |
| Noometry Index | 70.8 | 43.8 |
| Released | 2026-09-03 | 2026-04-17 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 1M |
| Max output | 128K | 30K |
| Input $ / M tokens | $10 | $1.25 |
| Output $ / M tokens | $50 | $2.50 |
| Results tracked | 56 | 40 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Astra leads
GPT-6 Astra: 73.7 (#2), Grok 4.3: 41.6 (#121)
| Benchmark | GPT-6 Astra | Grok 4.3 |
|---|---|---|
| LMArena WebDev | 1786 | 1357 |
| SciCode | 56.5% | 47.3% |
| WeirdML | 93.6% | 49.9% |
| LMArena Coding | 1487 | 1415 |
| ALE-Bench | 2,951 | 944.17 |
| DeepSWE | 74.1% | — |
| FrontierCode | 53.3% | — |
| FrontierSWE | 65.5% | — |
| GSO | 79.4% | — |
| MirrorCode | 46.7% | — |
Agentic & Tool Use GPT-6 Astra leads
GPT-6 Astra: 52.9 (#3), Grok 4.3: 27.7 (#99)
| Benchmark | GPT-6 Astra | Grok 4.3 |
|---|---|---|
| GDP.pdf | 34.2% | 8% |
| Vending-Bench 2 | 15,515 | 35.26 |
| APEX-Agents | 64.7% | — |
| Remote Labor Index | 20.8% | — |
| BALROG | 68.3% | — |
| LMArena Search | — | 1165 |
Reasoning GPT-6 Astra leads
GPT-6 Astra: 85.1 (#1), Grok 4.3: 35.9 (#68)
| Benchmark | GPT-6 Astra | Grok 4.3 |
|---|---|---|
| NYT Connections (extended) | 98.1% | 55.2% |
| CritPt | 31.7% | 8% |
| Chess Puzzles | 72% | 25% |
| LMArena Hard Prompts | 1462 | 1396 |
| DTBench | 97.3% | 90.7% |
| LMCA | 64.4% | 38.3% |
| Epoch Capabilities Index | 166.45 | 149.16 |
| ARC-AGI-2 | 95% | — |
| ARC-AGI-1 | 98.5% | — |
| EBR-Bench | 76.2% | — |
| Mystery Game Puzzles | 84% | — |
| Bench to the Future 3 | 0.14 | — |
| ForecastBench | — | 60.3 |
Math GPT-6 Astra leads
GPT-6 Astra: 93.5 (#2), Grok 4.3: 46.0 (#74)
| Benchmark | GPT-6 Astra | Grok 4.3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 93.7% | 42.8% |
| FrontierMath Tier 4 | 97.6% | 14.6% |
| OTIS Mock AIME 2024-2025 | 100% | 93.3% |
| ProofBench | 99% | 11% |
| LMArena Math | 1465 | 1388 |
| FrontierMath Erdős | 2.9% | — |
Knowledge GPT-6 Astra leads
GPT-6 Astra: 75.3 (#1), Grok 4.3: 52.5 (#62)
| Benchmark | GPT-6 Astra | Grok 4.3 |
|---|---|---|
| GPQA Diamond | 95.8% | 88.8% |
| SimpleQA Verified | 75.6% | 33.2% |
| LMArena Expert | 1483 | 1385 |
| Humanity's Last Exam | 54.8% | — |
| Vectara Hallucination Rate | 8.7% | — |
Multimodal GPT-6 Astra leads
GPT-6 Astra: 55.0 (#3), Grok 4.3: 31.6 (#104)
| Benchmark | GPT-6 Astra | Grok 4.3 |
|---|---|---|
| LMArena Vision | 1281 | 1229 |
| Blueprint-Bench 2 | 49.7% | 0% |
| Furniture Assembly | 80% | — |
| LMArena Document | 1468 | — |
Multilingual GPT-6 Astra leads
GPT-6 Astra: 53.7 (#61), Grok 4.3: 50.5 (#120)
| Benchmark | GPT-6 Astra | Grok 4.3 |
|---|---|---|
| LMArena Non-English | 1430 | 1385 |
| LMArena Chinese | 1484 | 1422 |
| LMArena French | 1456 | 1412 |
| LMArena German | 1440 | 1395 |
| LMArena Japanese | 1379 | 1379 |
| LMArena Korean | 1426 | 1356 |
| LMArena Russian | 1436 | 1399 |
| LMArena Spanish | 1407 | 1398 |
Instruction Following GPT-6 Astra leads
GPT-6 Astra: 76.3 (#44), Grok 4.3: 72.1 (#140)
| Benchmark | GPT-6 Astra | Grok 4.3 |
|---|---|---|
| LMArena Instruction Following | 1450 | 1366 |
Long Context GPT-6 Astra leads
GPT-6 Astra: 44.5 (#62), Grok 4.3: 42.5 (#123)
| Benchmark | GPT-6 Astra | Grok 4.3 |
|---|---|---|
| LMArena Longer Query | 1456 | 1393 |
Writing & Preference GPT-6 Astra leads
GPT-6 Astra: 75.3 (#7), Grok 4.3: 58.5 (#118)
| Benchmark | GPT-6 Astra | Grok 4.3 |
|---|---|---|
| LMArena Text | 1441 | 1397 |
| LMArena Creative Writing | 1418 | 1380 |
| LMArena Multi-Turn | 1448 | 1406 |
| EQ-Bench Creative Writing | 2173 | — |
| EQ-Bench 4 | — | 1075 |
Frequently asked questions
Is GPT-6 Astra better than Grok 4.3?
GPT-6 Astra is the stronger model overall, scoring 70.8 to 43.8 on the Noometry Index. Grok 4.3 costs 13× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Which is cheaper, GPT-6 Astra or Grok 4.3?
Grok 4.3 is cheaper. It lists at $1.25 per million input tokens and $2.50 per million output tokens; GPT-6 Astra lists at $10 and $50.
Is GPT-6 Astra or Grok 4.3 better for coding?
GPT-6 Astra scores higher on coding benchmarks: 73.7 versus 41.6 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Astra does, with 1.05M tokens against 1M.
How many benchmarks do GPT-6 Astra and Grok 4.3 share?
37 benchmarks have published results for both models. GPT-6 Astra has 56 scored results on Noometry and Grok 4.3 has 40.