Gemini 3.7 Flash vs Gemini 3.8 Flash
| Spec | Gemini 3.7 Flash | Gemini 3.8 Flash |
|---|---|---|
| Respan Index | 52.0 (#39) | 54.6 (#30) |
| Blended price per 1M | $1.5 | $1.5 |
| Context window | 1.0M | 1.0M |
| Input / output per 1M | $0.75 / $3.75 | $0.75 / $3.75 |
| Max output | 66K | 66K |
| Input | text, vision, audio | text, vision, audio |
| Released | 2026-08-13 | 2026-09-02 |
Respan Index by category
Each category is scored 0 to 100, and how far ahead the winner is. Leads under a point are close to even.
| Category | Gemini 3.7 Flash | Gemini 3.8 Flash |
|---|---|---|
| Coding | 32.4 Gemini 3.7 Flash | 36.1 Gemini 3.8 Flash |
| Agents | 66.1 Gemini 3.7 Flash | 66.7 Gemini 3.8 Flash |
| Reasoning | 48.1 Gemini 3.7 Flash | 53.3 Gemini 3.8 Flash |
| Knowledge & language | 69.2 Gemini 3.7 Flash | 70.9 Gemini 3.8 Flash |
| Human preference | 56.4 Gemini 3.7 Flash | 56.3 Gemini 3.8 Flash |
Benchmarks both report
Gemini 3.8 Flash leads on 17 of the 30 benchmarks both report. Higher is better. Each score links to where it was published.
| Benchmark | Gemini 3.7 Flash | Gemini 3.8 Flash |
|---|---|---|
| LMArena Agent (LMArena) | -0.0307 | 0.0149 |
| LMArena Elo | 1490 | 1497 |
| LMArena WebDev (LMArena) | 1588 | 1585 |
| LMArena Vision (LMArena) | 1294 | 1293 |
| MirrorCode (Epoch AI) | 41.1% | 38.9% |
| Chess Puzzles (Epoch AI) | 66% | 58% |
| APEX-Agents (Mercor) | 67.8% | 64.3% |
| DeepSWE v1.1 (Datacurve) | 65.5% | 73.8% |
| FrontierMath Tier 4 v2 (Epoch AI) | 36.6% | 22% |
| FrontierMath Tiers 1-3 v2 (Epoch AI) | 71.6% | 68.4% |
| FrontierSWE V2 (Proximal Labs) | 20.3% | 19.6% |
| Furniture Assembly (Epoch AI) | 26.7% | 31.7% |
| GPQA Diamond | 94.8% | 95.4% |
| LiveBench Agentic Coding (LiveBench) | 58.3% | 54.2% |
| LiveBench Coding (LiveBench) | 78.9% | 72.5% |
| LiveBench Data Analysis (LiveBench) | 68% | 54% |
| LiveBench Instruction Following (LiveBench) | 79.9% | 81.4% |
| LiveBench Language (LiveBench) | 85.5% | 87.8% |
| LiveBench Mathematics (LiveBench) | 93.5% | 91.6% |
| LiveBench Reasoning (LiveBench) | 87.8% | 89.3% |
| Mystery Game Puzzles (Epoch AI) | 37% | 47% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 97.2% | 98.9% |
| SAGE (Vals AI) | 49.2% | 35.1% |
| SimpleQA Verified | 69.2% | 69.7% |
| Terminal-Bench 4.0 (Vals AI) | 12.1% | 19.7% |
| DeepSWE v1.1 | 65.3% | 73.7% |
| GDPVal-AA v2 | 1525 | 1545 |
| HLE-Verified | 53.6% | 54.9% |
| OSWorld-2.0 | 47.9% | 59% |
| Terminal-bench 2.1 | 85.8% | 89.4% |
FAQ
Is Gemini 3.7 Flash better than Gemini 3.8 Flash?
Gemini 3.8 Flash scores 54.6 on the Respan Index (#30), ahead of Gemini 3.7 Flash at 52.0 (#39). Gemini 3.8 Flash leads on 17 of the 30 benchmarks both report.
Which is cheaper, Gemini 3.7 Flash or Gemini 3.8 Flash?
Gemini 3.7 Flash lists at $0.75 input and $3.75 output per 1M tokens; Gemini 3.8 Flash at $0.75 and $3.75. Both cost about the same per token at list price.
Which has the larger context window?
Both accept 1.0M tokens of context.
Compare other models
- vsGemini 3.7 Flash vs Gemini 3.6 Flash
- vsClaude Sonnet 5.5 vs Gemini 3.8 Flash
- vsGemini 3.8 Flash vs Claude Sonnet 5
- vsGemini 3.7 Flash vs GLM-5.3
- vsGemini 3.7 Flash vs MiMo-V2.6-Flash
- vsGemini 3.7 Flash vs MiniMax M3.1 Flash
- vsGemini 3.8 Flash vs GPT-5.5
- vsGemini 3.8 Flash vs Grok 4.6
- vsGemini 3.8 Flash vs GPT-5.6 Terra