token.app › Compare › GPT-5.6 Terra vs Gemini 3.6 Flash
GPT-5.6 Terra vs Gemini 3.6 Flash
GPT-5.6 Terra leads on 3 of 5 shared benchmarks. GPT-5.6 Terra is 1.3× cheaper per token on a blended 3:1 basis. GPT-5.6 Terra accepts the larger context window (1.1M).
Head to head
Blended price uses a 3:1 input:output mix
| GPT-5.6 Terra | Gemini 3.6 Flash | |
|---|---|---|
| Provider | OpenAI | |
| Input $/1M | $1.00 | $1.50 |
| Output $/1M | $6.00 | $7.50 |
| Blended $/1M | $2.25 | $3.00 |
| Context window | 1.1M | 1.0M |
| Max output | 128K | 66K |
| Released | Jul 2026 | Jul 2026 |
| Licence | Proprietary | Proprietary |
| GPQA Diamond | 93.3% | 94.1% |
| FrontierMath T1–3 | 86.0% | 58.9% |
| FrontierMath T4 | 70.7% | 22.0% |
| SimpleQA Verified | 43.1% | 68.7% |
| OTIS Mock AIME | 99.7% | 94.2% |
Benchmark scores independently run and published by Epoch AI, 'AI Benchmarking Hub', used under CC BY 4.0. Only benchmarks with a score for BOTH models are compared.
Choose GPT-5.6 Terra if…
- Costs less per token — $2.25 vs $3.00 blended
- Higher FrontierMath T1–3 (86.0% vs 58.9%)
- Higher FrontierMath T4 (70.7% vs 22.0%)
- Higher OTIS Mock AIME (99.7% vs 94.2%)
- Larger context window (1.1M)
Choose Gemini 3.6 Flash if…
- Higher GPQA Diamond (94.1% vs 93.3%)
- Higher SimpleQA Verified (68.7% vs 43.1%)
- Newer model (released Jul 2026)
FAQ
Which is better, GPT-5.6 Terra or Gemini 3.6 Flash?
Across the 5 benchmarks both models have independently-run scores for, GPT-5.6 Terra leads on 3 and Gemini 3.6 Flash leads on 2. Benchmarks are one input — price and context window are on this page too.
Is GPT-5.6 Terra cheaper than Gemini 3.6 Flash?
GPT-5.6 Terra is cheaper: $2.25 versus $3.00 per million tokens blended at 3:1 input:output. Input alone: $1.00 vs $1.50. Output alone: $6.00 vs $7.50.
What context windows do GPT-5.6 Terra and Gemini 3.6 Flash have?
GPT-5.6 Terra: 1,050,000 tokens. Gemini 3.6 Flash: 1,048,576 tokens.
Who makes GPT-5.6 Terra and Gemini 3.6 Flash?
GPT-5.6 Terra is made by OpenAI. Gemini 3.6 Flash is made by Google.