Gemini 4 Neon benchmark table shows broad lead over Claude Opus 5.5 and GPT-6.2
YashasGunderia · x · 2026-11-19
YashasGunderia posts the Gemini 4 Neon evaluation table against Claude Opus 5.5 and GPT-6.2 Sol, urging others to test it themselves. Gemini 4 Neon leads on most of the 18 benchmarks:
- Capability indices: AA Index v5.0.1 61 (vs 59/57), Vals Index 70.9%.
- Agentic coding: CursorBench 4.1 65.4% vs 58.1%; but it slightly trails Opus 5.5 on Terminal-bench 4.1 (65.2% vs 65.8%) and FrontierCode 2.0 (46.2% vs 47.3%); SWE-sweep provider harness: 39.0% vs 10.1%/8.3%.
- Long context: GraphWalks (256k–1M) 95.2% vs 66.8% for Opus.
- Reasoning/security: HLE-Diamond 71.3%, SEC-bench Pro 90.5%, WebExploitBench 51.3% — all leads.
Community reaction: wait for real-world testing before believing the scores.