Skip to content

Gemini 4 Neon vs GPT-6.2 Sol: Code Red for OpenAI?

Google has officially announced Gemini 4 Neon, just a day after OpenAI launched GPT-6.2 Sol. On Google's benchmarks, Neon beats GPT-6.2 Sol on 16 of 17 tests, and it leads the Artificial Analysis Intelligence Index 61.4 to 57.3. But Sol is still far cheaper to run, at about 84% less per task, and it finishes tasks in less than half the time. Here are the detailed benchmarks to help you see the real differences between these models.

Google's benchmarks

Benchmark
Gemini 4 Neon
GPT-6.2 Sol
Difference
AA Intelligence Index v5.0.16157Gemini 4 Neon+4
Vals Index v2.270.9%66.1%Gemini 4 Neon+4.8 pts
Terminal-Bench 4.165.2%61.5%Gemini 4 Neon+3.7 pts
FrontierCode 2.046.2%43.8%Gemini 4 Neon+2.4 pts
SWE-sweep (provider harness)39.0%8.3%Gemini 4 Neon+30.7 pts
SWE-sweep (Mini-SWE harness)12.8%11.6%Gemini 4 Neon+1.2 pts
AutomationBench56.3%41.3%Gemini 4 Neon+15.0 pts
Agents' Last Exam v1.175.8%73.0%Gemini 4 Neon+2.8 pts
RiemannBench84.0%80.0%Gemini 4 Neon+4.0 pts
Terminal-Bench Science 0.170.5%68.0%Gemini 4 Neon+2.5 pts
HLE-Diamond71.3%64.0%Gemini 4 Neon+7.3 pts
MKA (Climb WSR)56.5%52.7%Gemini 4 Neon+3.8 pts
Chartography73.5%75.1%GPT-6.2 Sol+1.6 pts
LVBench89.0%87.6%Gemini 4 Neon+1.4 pts
GraphWalks (256K to 1M)95.2%80.1%Gemini 4 Neon+15.1 pts
CWE-Bench v270.7%66.0%Gemini 4 Neon+4.7 pts
SEC-bench Pro90.5%88.3%Gemini 4 Neon+2.2 pts

Artificial Analysis benchmarks at the highest effort

Intelligence Index v5.0.161.457.3Gemini 4 Neon+4.1
AA-Briefcase v1.11700 Elo1612 EloGemini 4 Neon+88 Elo
AutomationBench-AA79.3%71.3%Gemini 4 Neon+8.0 pts
Terminal-Bench 4.063.6%60.1%Gemini 4 Neon+3.5 pts
FrontierCode-AA 2.044.4%44.4%
AA-Omniscience v1.154.840.4Gemini 4 Neon+14.4
GDP.pdf28.0%30.0%GPT-6.2 Sol+2.0 pts
HLE-Diamond71.3%64.0%Gemini 4 Neon+7.3 pts
CritPt v2 (Hard Subset)60.5%62.5%GPT-6.2 Sol+2.0 pts
Terminal-Bench Science 0.170.0%67.6%Gemini 4 Neon+2.4 pts

Speed, cost and tokens at the highest effort

Cost per Intelligence Index task$6.44$1.01GPT-6.2 Sol84% cheaper
Time per task21.5 min8.5 minGPT-6.2 Sol60% less time
Output tokens across the Index283M60.9MGPT-6.2 Sol78% fewer
Input tokens across the Index8.12B2.22BGPT-6.2 Sol73% fewer
Output speed110 tokens/s79 tokens/sGemini 4 Neon39% faster
Time to first answer token56.5 s75.1 sGemini 4 Neon25% lower

Pricing and specs

Input, per 1M tokens$3.00$2.00GPT-6.2 Sol33% cheaper
Output, per 1M tokens$15.00$10.00GPT-6.2 Sol33% cheaper
Cached input, per 1M tokens$0.15$0.10GPT-6.2 Sol33% cheaper
Cache writes, per 1M tokensImplicit caching only$2.50 for 30 min
Batch input, per 1M tokens$1.50$1.00GPT-6.2 Sol33% cheaper
Batch output, per 1M tokens$7.50$5.00GPT-6.2 Sol33% cheaper
Context window1.05M tokens1.05M tokens
Standard max output64K tokens128K tokensGPT-6.2 Sol2x larger
Extended outputUp to 1M with long decode128K tokens
Reasoning levelsLow, medium, high and xhighLow to maxGPT-6.2 SolAdds max
Text and image inputYesYes
Audio inputYesNo
Video inputYesNo
PDF inputYesYes
Computer useYesYes
Code executionYesYes
Free API tierNoNo
Knowledge cutoffMarch 2026June 1, 2026GPT-6.2 SolAbout 3 months newer
ReleasedNovember 18, 2026November 17, 2026Gemini 4 Neon1 day newer