Gemini 3.8 Flash vs Granite 4.2 8B
Specs, per-token pricing, and live benchmark scores — refreshed daily from our benchmark database.
Specs & pricing
| Gemini 3.8 Flash | Granite 4.2 8B | |
|---|---|---|
| Company | Google DeepMind | Ibm |
| Released | Sep 2026 | Aug 2026 |
| Context window | 1.0M | 131K |
| Input /1M tokens | $0.75 | $0.1 |
| Output /1M tokens | $3.75 | $0.15 |
| Open weights | No | No |
Benchmarks
| Benchmark | Gemini 3.8 Flash | Granite 4.2 8B |
|---|---|---|
| GPQA | — | 63.1% |
| MMLU-Pro | — | 74.0% |
| SWE-bench Verified | — | 47.7% |
| Speed (tok/s) | — | 153 tok/s |
Scores come from published leaderboards and official model cards (sources listed on the benchmarks page) and refresh daily. Pricing via vendor pages and OpenRouter.