Agnes AI has launched Agnes 2.5 Pro Alpha, a cost-effective reasoning model that scores 39 on the Artificial Analysis Intelligence Index, indicating a mid-pack performance among tested reasoning models. This model, which supports input from text, image, and video while providing text outputs, is available exclusively through Agnes AI’s first-party API. At a competitive price of $0.45 per million input tokens and $0.90 per million output tokens, Agnes 2.5 Pro Alpha is one of the lowest-priced models in its intelligence category, marking its debut with a focus on delivering efficient reasoning capabilities.

Agnes AI: Singapore-based AI lab that trains its own full-modality foundation models in-house across text, image, and video. It offers them through a free omni-modal API. Agnes AI has launched Agnes 2.5 Pro Alpha as its latest reasoning model emphasizing competitive performance at low cost.
GPT-5.4 mini: A proprietary reasoning model from OpenAI included in benchmark comparisons. It serves as a reference point for similarly priced models in intelligence, coding, and academic reasoning evaluations.
Qwen3.7 Plus: A reasoning model from the Qwen series featured in comparative analyses. It is benchmarked against Agnes 2.5 Pro Alpha on metrics such as coding performance and task-based reasoning.
DeepSeek V4 Flash: A competing reasoning model used as a benchmark in pricing and performance comparisons. It is referenced for its positioning among low-cost options in the same intelligence tier as Agnes 2.5 Pro Alpha.
Agnes 2.5 Pro Alpha: A text, image, and video input reasoning model with text output developed by Agnes AI. It marks the company’s debut on the Artificial Analysis Intelligence Index, where it places mid-pack among reasoning models. The model highlights coding as a relative strength while maintaining affordability.
Artificial Analysis Intelligence Index: A benchmark leaderboard evaluating AI models on intelligence metrics including coding, academic reasoning, and real-world tasks. Agnes 2.5 Pro Alpha made its first appearance here, establishing its position relative to other reasoning models.

Availability: The model is accessible exclusively through Agnes AI’s first-party API.
Multimodality: Agnes 2.5 Pro Alpha supports text, image, and video inputs with text outputs as a core feature of its design.
Benchmark Debut: Agnes 2.5 Pro Alpha enters the Artificial Analysis Intelligence Index as a new entrant focused on cost-efficient reasoning performance.