Model comparison
Gemini 3.8 Flash vs GPT-6 Astra
GPT-6 Astra is the stronger model overall, scoring 70.8 to 61.8 on the Noometry Index. Gemini 3.8 Flash 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 . 48 shared benchmarks.
GoogleSummary
- They share 48 benchmarks with published results for both. Gemini 3.8 Flash scores higher in 3 categories and GPT-6 Astra in 7 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-6 Astra leads 93.5 to 65.3.
- The biggest single-benchmark swing is FrontierMath Tier 4: 22% for Gemini 3.8 Flash and 97.6% for GPT-6 Astra.
- Gemini 3.8 Flash is cheaper at $0.75 / $3.75 per million input/output tokens, against $10 / $50 for GPT-6 Astra.
- GPT-6 Astra accepts more context: 1.05M tokens versus 1.05M.
Side by side
| Gemini 3.8 Flash | GPT-6 Astra | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 61.8 | 70.8 |
| Released | 2026-09-02 | 2026-09-03 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 1.05M |
| Max output | 66K | 128K |
| Input $ / M tokens | $0.75 | $10 |
| Output $ / M tokens | $3.75 | $50 |
| Results tracked | 50 | 56 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Astra leads
Gemini 3.8 Flash: 59.2 (#15), GPT-6 Astra: 73.7 (#2)
| Benchmark | Gemini 3.8 Flash | GPT-6 Astra |
|---|---|---|
| DeepSWE | 73.8% | 74.1% |
| FrontierCode | 41.2% | 53.3% |
| LMArena WebDev | 1584 | 1786 |
| FrontierSWE | 19.6% | 65.5% |
| SciCode | 56.6% | 56.5% |
| WeirdML | 84.8% | 93.6% |
| LMArena Coding | 1510 | 1487 |
| ALE-Bench | 1,270 | 2,951 |
| CursorBench | 39.6% | — |
| GSO | — | 79.4% |
| MirrorCode | — | 46.7% |
Agentic & Tool Use GPT-6 Astra leads
Gemini 3.8 Flash: 41.8 (#21), GPT-6 Astra: 52.9 (#3)
| Benchmark | Gemini 3.8 Flash | GPT-6 Astra |
|---|---|---|
| APEX-Agents | 64.3% | 64.7% |
| Remote Labor Index | 5.8% | 20.8% |
| GDP.pdf | 23.4% | 34.2% |
| Vending-Bench 2 | 5,094 | 15,515 |
| BALROG | — | 68.3% |
Reasoning GPT-6 Astra leads
Gemini 3.8 Flash: 76.9 (#5), GPT-6 Astra: 85.1 (#1)
| Benchmark | Gemini 3.8 Flash | GPT-6 Astra |
|---|---|---|
| ARC-AGI-2 | 89.2% | 95% |
| NYT Connections (extended) | 97.4% | 98.1% |
| ARC-AGI-1 | 98.5% | 98.5% |
| CritPt | 18.3% | 31.7% |
| Chess Puzzles | 61% | 72% |
| LMArena Hard Prompts | 1508 | 1462 |
| Mystery Game Puzzles | 47% | 84% |
| DTBench | 95.7% | 97.3% |
| LMCA | 52.9% | 64.4% |
| Epoch Capabilities Index | 156.71 | 166.45 |
| EBR-Bench | — | 76.2% |
| Surface Evolver Bench | 76.9% | — |
| Bench to the Future 3 | — | 0.14 |
Math GPT-6 Astra leads
Gemini 3.8 Flash: 65.3 (#28), GPT-6 Astra: 93.5 (#2)
| Benchmark | Gemini 3.8 Flash | GPT-6 Astra |
|---|---|---|
| FrontierMath (Tiers 1-3) | 68.4% | 93.7% |
| FrontierMath Tier 4 | 22% | 97.6% |
| OTIS Mock AIME 2024-2025 | 98.9% | 100% |
| ProofBench | 48% | 99% |
| LMArena Math | 1528 | 1465 |
| FrontierMath Erdős | — | 2.9% |
Knowledge Too close to call
Gemini 3.8 Flash: 74.8 (#2), GPT-6 Astra: 75.3 (#1)
| Benchmark | Gemini 3.8 Flash | GPT-6 Astra |
|---|---|---|
| GPQA Diamond | 95.4% | 95.8% |
| Humanity's Last Exam | 44.5% | 54.8% |
| SimpleQA Verified | 69.7% | 75.6% |
| LMArena Expert | 1524 | 1483 |
| Vectara Hallucination Rate | — | 8.7% |
Multimodal GPT-6 Astra leads
Gemini 3.8 Flash: 40.7 (#45), GPT-6 Astra: 55.0 (#3)
| Benchmark | Gemini 3.8 Flash | GPT-6 Astra |
|---|---|---|
| LMArena Vision | 1314 | 1281 |
| Blueprint-Bench 2 | 38.6% | 49.7% |
| Furniture Assembly | 31.7% | 80% |
| LMArena Document | — | 1468 |
Multilingual Gemini 3.8 Flash leads
Gemini 3.8 Flash: 58.0 (#5), GPT-6 Astra: 53.7 (#61)
| Benchmark | Gemini 3.8 Flash | GPT-6 Astra |
|---|---|---|
| LMArena Non-English | 1491 | 1430 |
| LMArena Chinese | 1554 | 1484 |
| LMArena French | 1498 | 1456 |
| LMArena German | 1493 | 1440 |
| LMArena Japanese | 1502 | 1379 |
| LMArena Korean | 1459 | 1426 |
| LMArena Russian | 1515 | 1436 |
| LMArena Spanish | 1485 | 1407 |
Instruction Following Gemini 3.8 Flash leads
Gemini 3.8 Flash: 78.0 (#13), GPT-6 Astra: 76.3 (#44)
| Benchmark | Gemini 3.8 Flash | GPT-6 Astra |
|---|---|---|
| LMArena Instruction Following | 1490 | 1450 |
Long Context Gemini 3.8 Flash leads
Gemini 3.8 Flash: 46.3 (#24), GPT-6 Astra: 44.5 (#62)
| Benchmark | Gemini 3.8 Flash | GPT-6 Astra |
|---|---|---|
| LMArena Longer Query | 1508 | 1456 |
Writing & Preference GPT-6 Astra leads
Gemini 3.8 Flash: 72.2 (#15), GPT-6 Astra: 75.3 (#7)
| Benchmark | Gemini 3.8 Flash | GPT-6 Astra |
|---|---|---|
| LMArena Text | 1499 | 1441 |
| LMArena Creative Writing | 1492 | 1418 |
| EQ-Bench Creative Writing | 1748 | 2173 |
| LMArena Multi-Turn | 1501 | 1448 |
Frequently asked questions
Is Gemini 3.8 Flash better than GPT-6 Astra?
GPT-6 Astra is the stronger model overall, scoring 70.8 to 61.8 on the Noometry Index. Gemini 3.8 Flash 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, Gemini 3.8 Flash or GPT-6 Astra?
Gemini 3.8 Flash is cheaper. It lists at $0.75 per million input tokens and $3.75 per million output tokens; GPT-6 Astra lists at $10 and $50.
Is Gemini 3.8 Flash or GPT-6 Astra better for coding?
GPT-6 Astra scores higher on coding benchmarks: 73.7 versus 59.2 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Astra does, with 1.05M tokens against 1.05M.
How many benchmarks do Gemini 3.8 Flash and GPT-6 Astra share?
48 benchmarks have published results for both models. Gemini 3.8 Flash has 50 scored results on Noometry and GPT-6 Astra has 56.