Model comparison
GPT-6 Astra vs o4-mini
GPT-6 Astra is the stronger model overall, scoring 70.3 to 41.6 on the Noometry Index. o4-mini costs 10× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Last verified . 32 shared benchmarks.
Summary
- They share 32 benchmarks with published results for both. GPT-6 Astra scores higher in 9 categories and o4-mini in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-6 Astra leads 85.1 to 24.6.
- The biggest single-benchmark swing is FrontierMath Tier 4: 97.6% for GPT-6 Astra and 4.9% for o4-mini.
- o4-mini is cheaper at $1.10 / $4.40 per million input/output tokens, against $10 / $50 for GPT-6 Astra.
- GPT-6 Astra accepts more context: 1.05M tokens versus 200K.
Side by side
| GPT-6 Astra | o4-mini | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 70.3 | 41.6 |
| Released | 2026-09-03 | 2025-04-16 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 200K |
| Max output | 128K | 100K |
| Input $ / M tokens | $10 | $1.10 |
| Output $ / M tokens | $50 | $4.40 |
| Results tracked | 51 | 60 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Astra leads
GPT-6 Astra: 70.5 (#3), o4-mini: 41.6 (#110)
| Benchmark | GPT-6 Astra | o4-mini |
|---|---|---|
| WeirdML | 93.6% | 52.6% |
| LMArena Coding | 1488 | 1369 |
| ALE-Bench | 2,951 | 826.17 |
| DeepSWE | 74.1% | — |
| FrontierCode | 53.3% | — |
| SWE-bench Verified (bash only) | — | 45% |
| Aider Polyglot | — | 72% |
| LMArena WebDev | 1792 | — |
| FrontierSWE | 65.5% | — |
| SciCode | 56.5% | — |
| GSO | — | 3.6% |
| MirrorCode | 46.7% | — |
| CadEval | — | 62% |
| AlgoTune | — | 1.72 |
Agentic & Tool Use GPT-6 Astra leads
GPT-6 Astra: 52.3 (#3), o4-mini: 32.3 (#57)
| Benchmark | GPT-6 Astra | o4-mini |
|---|---|---|
| APEX-Agents | 64.7% | — |
| Berkeley Function Calling Leaderboard | — | 53.2% |
| GDPval | — | 25.3% |
| Remote Labor Index | 20.8% | — |
| BALROG | 68.3% | — |
| METR Time Horizons | — | 63.9% |
| Vending-Bench 2 | 15,515 | — |
Reasoning GPT-6 Astra leads
GPT-6 Astra: 85.1 (#1), o4-mini: 24.6 (#143)
| Benchmark | GPT-6 Astra | o4-mini |
|---|---|---|
| ARC-AGI-2 | 95% | 6.1% |
| ARC-AGI-1 | 98.5% | 58.7% |
| CritPt | 31.7% | 0.6% |
| Chess Puzzles | 72% | 26% |
| LMArena Hard Prompts | 1464 | 1351 |
| Mystery Game Puzzles | 84% | 5% |
| DTBench | 97.3% | 77.6% |
| LMCA | 64.4% | 26.5% |
| Epoch Capabilities Index | 166.6 | 145.63 |
| SimpleBench | — | 38.7% |
| Kagi LLM Benchmark | — | 67.6% |
| NYT Connections (extended) | 98.1% | — |
| EnigmaEval | — | 9.2% |
| EBR-Bench | 76.2% | — |
| Bench to the Future 3 | 0.14 | — |
| ForecastBench | — | 61.8 |
Math GPT-6 Astra leads
GPT-6 Astra: 93.4 (#1), o4-mini: 40.2 (#79)
| Benchmark | GPT-6 Astra | o4-mini |
|---|---|---|
| FrontierMath (Tiers 1-3) | 93.7% | 36.1% |
| FrontierMath Tier 4 | 97.6% | 4.9% |
| OTIS Mock AIME 2024-2025 | 100% | 81.7% |
| LMArena Math | 1463 | 1390 |
| ProofBench | 99% | — |
| Omni-MATH | — | 72% |
| MATH Level 5 | — | 97.8% |
| FrontierMath (Feb 2025 set) | — | 24.8% |
| FrontierMath Erdős | 2.9% | — |
| FrontierMath Tier 4 (v1) | — | 6.3% |
Knowledge GPT-6 Astra leads
GPT-6 Astra: 75.3 (#1), o4-mini: 43.5 (#83)
| Benchmark | GPT-6 Astra | o4-mini |
|---|---|---|
| GPQA Diamond | 95.8% | 79.6% |
| Humanity's Last Exam | 54.8% | 18.1% |
| SimpleQA Verified | 75.6% | 19.6% |
| Vectara Hallucination Rate | 8.7% | 18.6% |
| LMArena Expert | 1486 | 1344 |
| MMLU-Pro | — | 82% |
| Confabulations | — | 15.8% |
| GPQA (HELM) | — | 73.5% |
Multimodal GPT-6 Astra leads
GPT-6 Astra: 55.1 (#3), o4-mini: 39.5 (#46)
| Benchmark | GPT-6 Astra | o4-mini |
|---|---|---|
| LMArena Vision | 1280 | 1194 |
| GeoBench | — | 64% |
| VPCT | — | 57.5% |
| Blueprint-Bench 2 | 49.7% | — |
| Furniture Assembly | 80% | — |
| LMArena Document | 1468 | — |
Multilingual GPT-6 Astra leads
GPT-6 Astra: 54.0 (#53), o4-mini: 47.2 (#149)
| Benchmark | GPT-6 Astra | o4-mini |
|---|---|---|
| LMArena Non-English | 1431 | 1336 |
| LMArena Chinese | 1495 | 1355 |
| LMArena French | 1467 | 1362 |
| LMArena Russian | 1434 | 1334 |
| LMArena Spanish | 1412 | 1346 |
| LMArena German | — | 1336 |
| LMArena Japanese | — | 1308 |
| LMArena Korean | — | 1313 |
Instruction Following GPT-6 Astra leads
GPT-6 Astra: 76.6 (#37), o4-mini: 75.4 (#63)
| Benchmark | GPT-6 Astra | o4-mini |
|---|---|---|
| LMArena Instruction Following | 1454 | 1321 |
| IFEval | — | 92.8% |
Long Context Too close to call
GPT-6 Astra: 44.7 (#57), o4-mini: 45.5 (#31)
| Benchmark | GPT-6 Astra | o4-mini |
|---|---|---|
| LMArena Longer Query | 1460 | 1315 |
| Fiction.LiveBench | — | 77.8% |
Writing & Preference GPT-6 Astra leads
GPT-6 Astra: 75.7 (#7), o4-mini: 54.2 (#148)
| Benchmark | GPT-6 Astra | o4-mini |
|---|---|---|
| LMArena Text | 1443 | 1353 |
| LMArena Creative Writing | 1423 | 1294 |
| LMArena Multi-Turn | 1457 | 1350 |
| Short-Story Creative Writing | — | 75% |
| EQ-Bench Creative Writing | 2173 | — |
| WildBench | — | 85.4% |
Frequently asked questions
Is GPT-6 Astra better than o4-mini?
GPT-6 Astra is the stronger model overall, scoring 70.3 to 41.6 on the Noometry Index. o4-mini costs 10× 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 o4-mini?
o4-mini is cheaper. It lists at $1.10 per million input tokens and $4.40 per million output tokens; GPT-6 Astra lists at $10 and $50.
Is GPT-6 Astra or o4-mini better for coding?
GPT-6 Astra scores higher on coding benchmarks: 70.5 versus 41.6 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Astra does, with 1.05M tokens against 200K.
How many benchmarks do GPT-6 Astra and o4-mini share?
32 benchmarks have published results for both models. GPT-6 Astra has 51 scored results on Noometry and o4-mini has 60.