Quick Take

Gemini 3.5 Flash (high) and Qwen3.7 Max both launched on May 19, 2026, marking a competitive day for AI development. Gemini 3.5 Flash (high) focuses on high-speed, measurable performance, while Qwen3.7 Max positions itself as a high-intelligence, cost-free alternative for developers.

Benchmark Read

Qwen3.7 Max holds a slight edge in the overall Intelligence index (56.6 vs 55.3) and a notable lead in the Coding index (50.1 vs 45). In specific benchmarks, Qwen3.7 Max outperforms Gemini in IFBench (0.805 vs 0.763) and TerminalBench Hard (0.507 vs 0.409). Conversely, Gemini 3.5 Flash (high) shows superior performance in HLE (0.41 vs 0.381), SciCode (0.531 vs 0.488), and TAU2 (0.953 vs 0.947). Both models show near-identical results in GPQA and LCR.

Cost and Speed

Gemini 3.5 Flash (high) features a clear pricing structure: $1.50/1M input and $9.00/1M output, resulting in a $3.38/1M blended rate. It also provides specific performance data, including an output speed of 233.027 tokens per second and a time to first token of 9.746 seconds. In contrast, Qwen3.7 Max is currently listed with no pricing (input/output/blended at $0.00/1M), though its specific output speed and latency metrics remain unknown.

Best Fit

Gemini 3.5 Flash (high) is best suited for production environments where latency and throughput are critical, as its performance metrics are fully documented. Qwen3.7 Max is an ideal candidate for developers looking for high-intelligence coding assistance without the financial burden of token-based pricing, provided they can accommodate the lack of published speed metrics.