Gemini 4 Neon vs Kimi K3.1 Pro
TryTop 2 now
Gemini 4 Neon leads by 5.6 capability points (≈95% interval +2.9 to +8.4): a clear lead over Kimi K3.1 Pro · P(higher) ≈ 100%
Wins 56 of 61 shared benchmarks (4 lost, 1 tied)
Capability vs. relative cost
Gemini 4 NeonKimi K3.1 ProClosed weightsOpen weights
Benchmark wins56 – 4 · 61 shared, 1 tied
Comparing matching setups only (32 benchmarks excluded) · suites counted once
Area← Gemini 4 NeonKimi K3.1 Pro →
Data: Vals AI, Surge AI, Arena and 10 othersatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Kimi K3.1 Pro minus Gemini 4 Neon.
| Mystery Game Puzzles | Games | 85.0 | 42.0 | −43.0 |
| Humanity's Sixth Sense | Perception & media › Spatial & visual reasoning | 75.9 | 33.2 | −42.7 |
| DAYJOB: Healthcare (October 2026 rescoring) | Health & medicine | 47.2 | 8.8 | −38.4 |
| Chess Puzzles (task version 2.0) | Games | 83.0 | 45.0 | −38.0 |
| DAYJOB: Accounting | Professional › Finance, accounting & tax | 49.6 | 13.6 | −36.0 |
| DAYJOB: Finance (October 2026 rescoring) | Professional › Finance, accounting & tax | 50.5 | 17.0 | −33.5 |
| Furniture Assembly Benchmark | Perception & media › Spatial & visual reasoning | 89.2 | 55.8 | −33.3 |
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 71.3 | 39.4 | −31.9 |
| FrontierMath Tier 4 (v2) - Private | Math | 97.6 | 65.9 | −31.7 |
| Legal Research Bench Constitutional/Civil Rights | Professional › Legal | 63.2 | 31.6 | −31.6 |
| Terminal-Bench Science 0.1 (Vals) | Science › Multidisciplinary | 62.9 | 31.4 | −31.4 |
| GDP.xlsx | Professional › General knowledge work | 43.1 | 14.0 | −29.1 |
| LMArena Agent Arena Net Improvement (Steering Burden) | Tool & computer use › General agents | 18.4 | 3.9 | −28.3 |
| Legal Research Bench Business/Commercial | Professional › Legal | 55.8 | 27.9 | −27.9 |
| ExploitGym - release-reported percentage, cohort unspecified | Software › Cybersecurity | 48.3 | 20.7 | −27.6 |
| Legal Research Bench Immigration | Professional › Legal | 45.5 | 18.2 | −27.3 |
| Bug Hunt Bench | Software › Software engineering | 60.0 | 34.3 | −25.7 |
| CWE-bench v2 Pass@1 | Software › Cybersecurity | 70.7 | 45.3 | −25.4 |
| Legal Research Bench Criminal | Professional › Legal | 52.6 | 28.1 | −24.6 |
| RiemannBench (November 2026 rescoring) | Math | 84.0 | 60.8 | −23.2 |
| Legal Research Bench Family | Professional › Legal | 36.4 | 13.6 | −22.7 |
| sudo L7 (October 2026 rescoring) | Software › Software engineering | 38.3 | 16.7 | −21.6 |
| Legal Research Bench Weighted Pass | Professional › Legal | 91.6 | 71.4 | −20.1 |
| MysteryMechanism | Math | 45.0 | 26.1 | −18.9 |
| LMArena Agent Arena Praise vs Complaint | Tool & computer use › General agents | 44.5 | 6.7 | −18.9 |
| LiveBench 2.0 (2026-11-06 release) | Composite indexes | 87.4 | 68.7 | −18.8 |
| Legal Research Bench Civil Litigation | Professional › Legal | 48.1 | 29.6 | −18.5 |
| Vals Science Index | Science › Multidisciplinary | 62.6 | 44.4 | −18.3 |
| Code Migration v1.1 (Vals Index v2.2 subset) | Software › Software engineering | 75.8 | 59.7 | −16.1 |
| Tax Agent Bench | Professional › Finance, accounting & tax | 79.8 | 63.9 | −15.9 |
| Legal Research Bench Administrative/Regulatory | Professional › Legal | 71.2 | 56.1 | −15.2 |
| APEX-Agents 1.1 | Professional › General knowledge work | 88.9 | 75.9 | −13.0 |
| SimpleQA Verified (Epoch anti-abstention) | Knowledge & reasoning › General knowledge | 75.2 | 62.2 | −13.0 |
| FrontierMath Tiers 1-3 (v2) - Private | Math | 91.2 | 78.2 | −13.0 |
| Harvey LAB (Mercor run, undisclosed ~213-task cohort) | Professional › Legal | 26.6 | 13.8 | −12.8 |
| Vals Finance Index | Professional › Finance, accounting & tax | 68.3 | 56.0 | −12.3 |
| EnterpriseBench: CoreCraft | Tool & computer use › Customer service & operations | 77.4 | 66.1 | −11.3 |
| DAYJOB: Tax | Professional › Finance, accounting & tax | 12.8 | 1.6 | −11.2 |
| LMArena Code Arena WebDev | Software › Frontend & design | 1764 | 1684 | −10.6 |
| APEX-1 (Mercor, mean score) | Professional › General knowledge work | 79.4 | 69.3 | −10.1 |
| Artificial Analysis Coding Agent Index v2.1 | Software › Software engineering | 58.6 | 48.6 | −10.0 |
| Legal Research Bench Health | Professional › Legal | 83.3 | 73.3 | −10.0 |
| Vibe Code Bench v1.1 | Software › Frontend & design | 93.7 | 84.9 | −8.8 |
| LMArena Agent Arena Confirmed Success | Tool & computer use › General agents | 19.2 | 1.9 | −8.6 |
| APEX-Accounting 1.1 | Professional › Finance, accounting & tax | 70.8 | 62.4 | −8.4 |
| LMArena Agent Arena Steering Burden | Tool & computer use › General agents | 21.2 | 4.5 | −8.3 |
| HLAB-Vals-1.1 | Professional › Legal | 19.6 | 11.7 | −7.9 |
| Finance Agent v2.1 | Professional › Finance, accounting & tax | 68.6 | 61.5 | −7.1 |
| GDP.pdf (November 2026 rescoring) | Professional › General knowledge work | 53.4 | 46.6 | −6.8 |
| Excel Modeling Benchmark v1.1 | Professional › Finance, accounting & tax | 76.5 | 70.0 | −6.5 |
| LMArena Agent Arena Bash Reliability | Tool & computer use › General agents | -6.0 | 6.2 | +6.1 |
| AutomationBench-AA 1.0.6 | Tool & computer use › Customer service & operations | 79.3 | 73.3 | −6.0 |
| LMArena Text Arena Elo | Language & writing › General assistance | 1518 | 1487 | −5.6 |
| BioMysteryBench (Vals Terminus 2) | Science › Life sciences | 80.0 | 75.6 | −4.4 |
| DAYJOB: Law (November 2026 release) | Professional › Legal | 26.2 | 22.0 | −4.2 |
| Terminal-Bench 4.0 | Software › Software engineering | 61.6 | 63.6 | +2.0 |
| Agents' Last Exam v1.1 Score | Professional › General knowledge work | 75.8 | 74.4 | −1.4 |
| AA-Briefcase v1.1 Elo | Professional › General knowledge work | 1699 | 1678 | −1.2 |
| τ³-Banking (Mercor run, bm25_grep retrieval) | Tool & computer use › Customer service & operations | 59.5 | 59.8 | +0.3 |
| LMArena Agent Arena Tool Hallucination | Tool & computer use › Tool calling | 1.0 | 1.2 | +0.1 |
| OTIS Mock AIME 2024-2025 | Math | 100.0 | 100.0 | 0.0 |
Data: Vals AI, Surge AI, Arena and 10 othersatlas.village.dev