GPT-5.6 Sol Pro vs GPT-5.6 Terra
GPT-5.6 Sol Pro leads by 2.9 capability points (≈95% interval +0.5 to +5.3): a clear lead over GPT-5.6 Terra · P(higher) ≈ 99%
Wins 13 of 21 shared benchmarks (7 lost, 1 tied)
Each line runs through one model’s priced reasoning settings, cheapest to most expensive; grey points are other models’ priced settings in the same range. Results whose reasoning effort the source didn’t report are left out wherever a model has named effort levels. Capability and relative cost are the published estimates (relative cost is cost per task as a multiple of GPT-5 (high)); hover a setting, or use the arrow keys, to set it beside the other model’s same setting. GPT-5.6 Sol Pro has no setting with a relative cost yet.
GPT-5.6 Sol ProvsGPT-5.6 Terracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev
Benchmark wins13 – 7 · 21 shared, 1 tied
GPT-5.6 Sol ProvsGPT-5.6 Terrawins by categoryData Nov 15, 2026Data: Mercor, Epoch AI, Eval Connections and 7 othersVillage LLM Atlasatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-5.6 Terra minus GPT-5.6 Sol Pro.
| NanoGPT Speedrun Frontier (human record gap closed) | Agents | 33.6 | 11.0 | −30.4 |
| SWE-bench Verified Extended (Mercor held-out) | Code | 63.3 | 78.7 | +15.4 |
| SWE Atlas - Codebase QnA | Code | 30.9 | 45.2 | +14.3 |
| APEX-Accounting | Finance & Accounting | 51.5 | 40.3 | −11.2 |
| WeirdML (v2) | Science | 89.4 | 78.3 | −11.2 |
| BrowseComp (Mercor 130-question subset) | Search | 95.7 | 85.8 | −9.9 |
| FrontierMath Tier 4 (v2) - Private | Math | 80.5 | 70.7 | −9.8 |
| SWE Atlas Codebase QnA Extended (Mercor held-out) | Code | 24.7 | 16.0 | −8.7 |
| GeneBench-Pro | Science | 31.5 | 23.3 | −8.2 |
| Terminal-Bench 2.1 | Code | 81.2 | 88.0 | +6.8 |
| APEX-Agents 1.1 | Professional Work | 65.2 | 71.8 | +6.6 |
| Terminal-Bench 4.0 | Code | 32.3 | 26.3 | −6.0 |
| GPQA Diamond | General QA | 94.9 | 89.9 | −5.0 |
| SWE-bench Verified | Code | 72.8 | 77.8 | +5.0 |
| τ³-Banking (Mercor run, bm25_grep retrieval) | Agents | 43.0 | 38.8 | −4.2 |
| APEX-1 (Mercor, mean score) | Professional Work | 73.1 | 69.5 | −3.6 |
| WordleBench | Games | 94.0 | 91.0 | −3.0 |
| RedlineBench | Legal | 45.8 | 47.8 | +2.0 |
| Ruby LLM Benchmarks — Program Fixer | Code | 75.3 | 74.0 | −1.3 |
| SWE-bench Multilingual | Code | 94.2 | 94.4 | +0.2 |
| Eval Connections | Games | 100.0 | 100.0 | 0.0 |
GPT-5.6 Sol ProvsGPT-5.6 Terranormalised score marginsData Nov 15, 2026Data: Mercor, Epoch AI, Eval Connections and 7 othersVillage LLM Atlasatlas.village.dev