Benchmark scores
Higher is better for all benchmark scores (SWE-bench, GPQA, HLE, Terminal-Bench, etc.). Green marks the model with the higher score.
Side-by-side Gemini 4 Hydrogen vs Gemini 3.1 Pro comparison across SWE-bench, GPQA, HLE, Terminal-Bench, coding agent scores, token pricing, context window, and AskClash RWT. Green marks the winner on each benchmark.
Verdict: Gemini 4 Hydrogen ranks higher (#13 vs #46); Gemini 4 Hydrogen costs less ($1.50/$7.50 vs $2.00/$12.0 per 1M tokens).
Green cells highlight the winning model for each metric. Scores are cached from the AskClash LLM leaderboard snapshot.
| Metric | Gemini 4 Hydrogen | Gemini 3.1 Pro |
|---|---|---|
| Overall Score | 62.8 | 34.3 |
| Leaderboard Rank | #13 | #46 |
| ACB | 70.2 | — |
| AA Intelligence Index | 53.9 | — |
| AutomationBench-AA | 76.1 | — |
| HLE | — | 51.4 |
| GPQA | — | 94.3 |
| SWE-bench | — | 80.6 |
| SWE-Pro | — | 54.2 |
| Terminal-Bench 4.0 | 54.0 | — |
| DeepSWE | — | 11.7 |
| GDPval-AA | — | 1317.0 |
| MCP Atlas | — | 69.2 |
| Finance Agent v2.1 | 65.1 | — |
| CharXiv | — | 80.2 |
| MMMU-Pro | — | 83.9 |
| ARC-AGI 2 | — | 77.1 |
| Tau2 | — | 99.3 |
| MRCR | — | 84.9 |
| Input Price (per 1M tokens) | $1.50 | $2.00 |
| Output Price (per 1M tokens) | $7.50 | $12.0 |
| Context Window | 1M | 1M |
| Benchmarks Published | 5 | 9 |
Gemini 4 Hydrogen: its diagnostic ACB uses a review slice and sampled video frames.
Gemini 4 Hydrogen and Gemini 3.1 Pro do not fully overlap on published benchmark scores in this snapshot. Gemini 4 Hydrogen is cheaper on a defined 1M-input / 200K-output workload. Charts show only benchmarks both models publish.
Workload = published cost of 1M input + 200K output tokens. Open the live leaderboard for interactive compare charts.
Explore how Gemini 4 Hydrogen and Gemini 3.1 Pro stack up against other top-ranked LLMs.
Higher is better for all benchmark scores (SWE-bench, GPQA, HLE, Terminal-Bench, etc.). Green marks the model with the higher score.
Lower is better for input and output prices. Green marks the cheaper model per 1M tokens.
Models with fewer disclosed benchmark cells may have inflated percentile scores. Check the benchmark cell count for context.
This comparison page is generated from the AskClash LLM leaderboard cache. Open the live leaderboard for real-time scores and interactive filtering.