Up to four models
GPT-6 Astra vs Qwen3.8-27B
On the index (GPT-4 = 100), GPT-6 Astra scores 193 and Qwen3.8-27B 155. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
GPT-6 Astra
193178–209
WeightsClosed
ReleasedSep 2026
Price in / out$10 / $50
Context1.05M
Core tests taken7 of 9
155139–171
WeightsOpen
ReleasedAug 2026
Price in / out$0.060 / $2.2
Context1.02M
Core tests taken3 of 9
GPT-6 Astra leads Qwen3.8-27B by 38 points, ahead on 3 of 3 shared core tests, at 34× the price.
Test by testthird-party results
GPT-6 Astra
Qwen3.8-27B
Every result as a table
| Test | GPT-6 Astra | Qwen3.8-27B |
|---|---|---|
| Reasoning & knowledge | ||
| LiveBench Data Analysis | 83.0% | 76.6% |
| LiveBench IF | 75.6% | 72.7% |
| LiveBench Language | 89.4% | 74.4% |
| LiveBench Reasoning | 92.7% | 80.0% |
| SimpleBench | 83.6% | 60.2% |
| Maths | ||
| LiveBench Mathematics | 96.8% | 86.2% |
| ProofBench v1.1 | 99.0% | 16.0% |
| Coding | ||
| LiveBench Agentic Coding | 57.3% | 61.4% |
| LiveBench Coding | 80.4% | 75.7% |
| Code Migration | 67.7% | 14.2% |
| CyberBench v1.1 | 41.1% | 67.8% |
| IOI | 100.0% | 39.1% |
| ProgramBench | 5.5% | 0.0% |
| Vibe Code Bench v1.1 | 89.6% | 64.8% |
| Agents | ||
| APEX-Agents | 64.7% | 47.5% |
| Terminal-Bench-Science | 62.9% | 0.0% |
| DTBench | 97.3% | 88.0% |
| LMCA | 64.4% | 41.4% |
| Professional work | ||
| Finance Agent (v2) | 53.5% | 48.5% |
| EMB | 71.7% | 59.7% |
| Legal Research Bench | 39.4% | 36.1% |
| MedCode | 48.5% | 28.7% |
| MedScribe | 87.9% | 83.9% |
| Tax Agent Bench | 20.7% | 30.9% |
| Novel problems | ||
| ARC-AGI-2 | 95.0% | 42.4% |
| ARC-AGI-1 | 98.5% | 87.5% |
Tests at least two of these models have taken. Bold rows are the core tests.