Model comparison
GPT-4o vs GPT-6 Astra
GPT-6 Astra is the stronger model overall, scoring 70.8 to 28.6 on the Noometry Index. GPT-4o costs 4.6× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Last verified . 34 shared benchmarks.
Summary
- They share 34 benchmarks with published results for both. GPT-4o scores higher in 0 categories and GPT-6 Astra in 10 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-6 Astra leads 93.5 to 10.6.
- The biggest single-benchmark swing is ARC-AGI-2: 0% for GPT-4o and 95% for GPT-6 Astra.
- GPT-4o is cheaper at $2.50 / $10 per million input/output tokens, against $10 / $50 for GPT-6 Astra.
- GPT-6 Astra accepts more context: 1.05M tokens versus 128K.
Side by side
| GPT-4o | GPT-6 Astra | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 28.6 | 70.8 |
| Released | 2024-05-13 | 2026-09-03 |
| Weights | Proprietary | Proprietary |
| Context window | 128K | 1.05M |
| Max output | 16K | 128K |
| Input $ / M tokens | $2.50 | $10 |
| Output $ / M tokens | $10 | $50 |
| Results tracked | 72 | 56 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Astra leads
GPT-4o: 24.8 (#328), GPT-6 Astra: 73.7 (#2)
| Benchmark | GPT-4o | GPT-6 Astra |
|---|---|---|
| GSO | 0% | 79.4% |
| WeirdML | 25.1% | 93.6% |
| LMArena Coding | 1297 | 1487 |
| SWE-bench Verified | 31% | — |
| DeepSWE | — | 74.1% |
| FrontierCode | — | 53.3% |
| SWE-bench Verified (bash only) | 21.6% | — |
| Aider Polyglot | 45.3% | — |
| LMArena WebDev | — | 1786 |
| FrontierSWE | — | 65.5% |
| SciCode | — | 56.5% |
| BigCodeBench Instruct | 51.1% | — |
| LiveBench Coding | 51.4% | — |
| MirrorCode | — | 46.7% |
| BigCodeBench Complete | 61.1% | — |
| CadEval | 26% | — |
| ALE-Bench | — | 2,951 |
| HumanEval+ | 87.2% | — |
| MBPP+ | 72.2% | — |
Agentic & Tool Use GPT-6 Astra leads
GPT-4o: 21.0 (#141), GPT-6 Astra: 52.9 (#3)
| Benchmark | GPT-4o | GPT-6 Astra |
|---|---|---|
| BALROG | 32.3% | 68.3% |
| APEX-Agents | — | 64.7% |
| GDPval | 9.9% | — |
| Remote Labor Index | — | 20.8% |
| TheAgentCompany | 8.6% | — |
| Cybench | 12.5% | — |
| GDP.pdf | — | 34.2% |
| LMArena Search | 1006 | — |
| METR Time Horizons | 40.8% | — |
| Vending-Bench 2 | — | 15,515 |
Reasoning GPT-6 Astra leads
GPT-4o: 9.4 (#343), GPT-6 Astra: 85.1 (#1)
| Benchmark | GPT-4o | GPT-6 Astra |
|---|---|---|
| ARC-AGI-2 | 0% | 95% |
| ARC-AGI-1 | 4.5% | 98.5% |
| CritPt | 0% | 31.7% |
| Chess Puzzles | 13% | 72% |
| LMArena Hard Prompts | 1281 | 1462 |
| DTBench | 64.5% | 97.3% |
| LMCA | 16.6% | 64.4% |
| Epoch Capabilities Index | 128.97 | 166.45 |
| SimpleBench | 17.8% | — |
| NYT Connections (extended) | — | 98.1% |
| EnigmaEval | 0.8% | — |
| EBR-Bench | — | 76.2% |
| LiveBench Reasoning | 55.8% | — |
| Mystery Game Puzzles | — | 84% |
| LiveBench Data Analysis | 60.9% | — |
| Bench to the Future 3 | — | 0.14 |
| ForecastBench | 57.7 | — |
| LiveBench | 55.3% | — |
Math GPT-6 Astra leads
GPT-4o: 10.6 (#312), GPT-6 Astra: 93.5 (#2)
| Benchmark | GPT-4o | GPT-6 Astra |
|---|---|---|
| FrontierMath (Tiers 1-3) | 0.4% | 93.7% |
| OTIS Mock AIME 2024-2025 | 6.4% | 100% |
| LMArena Math | 1285 | 1465 |
| FrontierMath Tier 4 | — | 97.6% |
| ProofBench | — | 99% |
| Omni-MATH | 29.3% | — |
| LiveBench Math | 49.5% | — |
| MATH Level 5 | 53.3% | — |
| FrontierMath (Feb 2025 set) | 0.3% | — |
| FrontierMath Erdős | — | 2.9% |
Knowledge GPT-6 Astra leads
GPT-4o: 28.8 (#242), GPT-6 Astra: 75.3 (#1)
| Benchmark | GPT-4o | GPT-6 Astra |
|---|---|---|
| GPQA Diamond | 49.2% | 95.8% |
| Humanity's Last Exam | 2.7% | 54.8% |
| SimpleQA Verified | 26% | 75.6% |
| Vectara Hallucination Rate | 9.6% | 8.7% |
| LMArena Expert | 1250 | 1483 |
| MMLU-Pro | 71.3% | — |
| Confabulations | 15.3% | — |
| GPQA (HELM) | 52% | — |
| MMLU | 88.1% | — |
Multimodal GPT-6 Astra leads
GPT-4o: 34.5 (#91), GPT-6 Astra: 55.0 (#3)
| Benchmark | GPT-4o | GPT-6 Astra |
|---|---|---|
| LMArena Vision | 1137 | 1281 |
| Video-MME | 71.9% | — |
| GeoBench | 71% | — |
| VPCT | 40% | — |
| Blueprint-Bench 2 | — | 49.7% |
| Furniture Assembly | — | 80% |
| LMArena Document | — | 1468 |
| ScienceQA | 88.5% | — |
Multilingual GPT-6 Astra leads
GPT-4o: 43.2 (#186), GPT-6 Astra: 53.7 (#61)
| Benchmark | GPT-4o | GPT-6 Astra |
|---|---|---|
| LMArena Non-English | 1283 | 1430 |
| LMArena Chinese | 1277 | 1484 |
| LMArena French | 1304 | 1456 |
| LMArena German | 1282 | 1440 |
| LMArena Japanese | 1257 | 1379 |
| LMArena Korean | 1234 | 1426 |
| LMArena Russian | 1286 | 1436 |
| LMArena Spanish | 1292 | 1407 |
Instruction Following GPT-6 Astra leads
GPT-4o: 66.6 (#207), GPT-6 Astra: 76.3 (#44)
| Benchmark | GPT-4o | GPT-6 Astra |
|---|---|---|
| LMArena Instruction Following | 1278 | 1450 |
| LiveBench Instruction Following | 68.6% | — |
| IFEval | 81.7% | — |
Long Context GPT-6 Astra leads
GPT-4o: 39.4 (#179), GPT-6 Astra: 44.5 (#62)
| Benchmark | GPT-4o | GPT-6 Astra |
|---|---|---|
| LMArena Longer Query | 1289 | 1456 |
| Fiction.LiveBench | 66.7% | — |
Writing & Preference GPT-6 Astra leads
GPT-4o: 52.6 (#166), GPT-6 Astra: 75.3 (#7)
| Benchmark | GPT-4o | GPT-6 Astra |
|---|---|---|
| LMArena Text | 1300 | 1441 |
| LMArena Creative Writing | 1292 | 1418 |
| LMArena Multi-Turn | 1302 | 1448 |
| Short-Story Creative Writing | 81.8% | — |
| EQ-Bench Creative Writing | — | 2173 |
| WildBench | 82.8% | — |
| LiveBench Language | 47.6% | — |
Frequently asked questions
Is GPT-4o better than GPT-6 Astra?
GPT-6 Astra is the stronger model overall, scoring 70.8 to 28.6 on the Noometry Index. GPT-4o costs 4.6× 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-4o or GPT-6 Astra?
GPT-4o is cheaper. It lists at $2.50 per million input tokens and $10 per million output tokens; GPT-6 Astra lists at $10 and $50.
Is GPT-4o or GPT-6 Astra better for coding?
GPT-6 Astra scores higher on coding benchmarks: 73.7 versus 24.8 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Astra does, with 1.05M tokens against 128K.
How many benchmarks do GPT-4o and GPT-6 Astra share?
34 benchmarks have published results for both models. GPT-4o has 72 scored results on Noometry and GPT-6 Astra has 56.