Compare / head-to-head
Gemma 4 31BvsMAI-Thinking-1
Gemma 4 31B and MAI-Thinking-1 split their 4 shared benchmarks evenly. Both offer a 256K-token context window.
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
2 – 2
even split
Cheaper per token
—
list price, input + output
Larger context
Tie
both 256K tokens
Provider uptime (30d)
100% · —
Google · Microsoft
Gemma 4 31B
Google · released 2026-04-02
textvision
- Context
- 256K
- Max out
- —
- Input /1M
- —
- Output /1M
- —
- Cached /1M
- —
- Scores
- 6 · 5 core
MAI-Thinking-1
Microsoft · released 2026-08-12
text
- Context
- 256K
- Max out
- 64K
- Input /1M
- —
- Output /1M
- —
- Cached /1M
- —
- Scores
- 15 · 7 core
Quality
Benchmark matrix
4 shared · 2 only Gemma 4 31B · 11 only MAI-Thinking-1| Benchmark | Gemma 4 31B | MAI-Thinking-1 | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 4 | ||||
| GPQA Diamond | 84.3% ↗ | 84.2% ↗ | +0.1 pt | Gemma 4 31B |
| MMLU-Pro | 85.2% ↗ | 85% ↗ | +0.2 pt | Gemma 4 31B |
| AIME 2026 | 89.2% ↗ | 94.5% ↗ | -5.3 pt | MAI-Thinking-1 |
| LiveCodeBench v6 | 80% ↗ | 87.7% ↗ | -7.7 pt | MAI-Thinking-1 |
| Only Gemma 4 31B reports · 2 | ||||
| Humanity's Last Exam (no tools) | 19.5% ↗ | not reported | — | — |
| LMArena Elo | 1441.7 ↗ | not reported | — | — |
| Only MAI-Thinking-1 reports · 11 | ||||
| SWE-bench Verified | not reported | 73.5% ↗ | — | — |
| AIME 2025 | not reported | 97% ↗ | — | — |
| AdvancedIF | not reported | 85% ↗ | — | — |
| BFCL v3 | not reported | 72% ↗ | — | — |
| GraphWalks (<=128K) | not reported | 90% ↗ | — | — |
| HMMT February 2026 | not reported | 84.9% ↗ | — | — |
| IFBench | not reported | 69% ↗ | — | — |
| MultiChallenge | not reported | 53% ↗ | — | — |
| SimpleQA Verified | not reported | 31% ↗ | — | — |
| SWE-bench Pro | not reported | 52.8% ↗ | — | — |
| Terminal-Bench 2.0 | not reported | 46% ↗ | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is Gemma 4 31B minus MAI-Thinking-1 in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | Gemma 4 31B | MAI-Thinking-1 | Edge |
|---|---|---|---|
| Context window | 256K tokens | 256K tokens | Tie |
| Max output | — | 64K tokens | — |
| Input price / 1M | — | — | — |
| Output price / 1M | — | — | — |
| Cached input / 1M | — | — | — |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | — | — | — |
| Modalities | text · vision | text | Gemma 4 31B |
| Released | 2026-04-02 | 2026-08-12 | — |
| Cited benchmark scores | 6 | 15 | — |
Reliability
Provider status
Last 60 days · refreshed daily on this page · live on /statusMore matchups
Gemma 4 31B vs …
Models sharing the most benchmarksMore matchups
MAI-Thinking-1 vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run Gemma 4 31B and MAI-Thinking-1 on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.