Gemini 3.6 Flash (High) Scores 50 on Artificial Analysis Intelligence Index, Ranks Among Top-Tier Models
Key Takeaways
- ▸Gemini 3.6 Flash (high) scored 50 on the Artificial Analysis Intelligence Index v4.1, significantly outperforming the category average of 31
- ▸The model supports multimodal inputs (text, image, speech, video) with a 1M token context window and reasoning capabilities
- ▸Pricing is competitive: $1.50/1M input tokens and $7.50/1M output tokens, with cache hits receiving a 90% discount
Summary
Google's Gemini 3.6 Flash (high) has been evaluated on the Artificial Analysis Intelligence Index v4.1 and achieved a score of 50, placing it well above the average benchmark score of 31 and among the leading models in its class. The reasoning-capable model supports multimodal inputs (text, image, speech, and video), offers text output, and features a 1 million token context window—equivalent to approximately 1,500 pages of text.
In terms of pricing competitiveness, Gemini 3.6 Flash (high) is moderately positioned: input tokens cost $1.50 per 1M (compared to the category average of $1.75) and output tokens cost $7.50 per 1M (average: $9.00). Cache hits receive an aggressive 90% discount at $0.15 per 1M tokens. The model demonstrated efficiency during evaluation, generating 59M tokens—below the category average of 63M—while achieving high intelligence scores.
The benchmark evaluation, conducted across 22 intelligence evaluations including agentic real-world tasks, tool use, coding, scientific reasoning, and long-context reasoning, positions Gemini 3.6 Flash (high) as a competitive option for enterprise and general-purpose LLM workloads. Google's multimodal capabilities and reasoning features differentiate the model in an increasingly crowded market of high-performance language models.
- Achieved strong efficiency metrics, generating 59M tokens during evaluation—below the category average—while maintaining high intelligence scores
Editorial Opinion
Google's Gemini 3.6 Flash (high) demonstrates that the company remains competitive in the high-performance LLM space, particularly on the intelligence-to-price metric that matters most to enterprise buyers. The multimodal capabilities and reasoning features position it as a practical choice for complex agentic tasks, though pricing remains slightly premium compared to competitors like Claude 3.5 Sonnet. The aggressive cache hit discount signals Google's confidence in long-context applications and suggests a strategic focus on sustained engagement over one-shot queries. This benchmark reinforces Google's commitment to offering multiple Gemini variants optimized for different performance tiers and use cases.
