Best Open-Source AI Models
This ranked list highlights leading open-weight models suited for large-scale text, code, and multimodal workloads. Key considerations include context window size, intelligence_index scores where available, output speed, per-million-token pricing, supported modalities, and documented strengths versus limitations such as text-only constraints or hallucination risks. All entries are open-weight with provider details drawn directly from the model specifications.
It ranks second due to its 51.5 intelligence_index, 79.81 t/s speed, $0.87 per million tokens cost, and 1048576-token context, excelling in coding tasks, long inputs, and technical domains.
It earns the top spot through its 46.5 intelligence_index, 103.73 t/s output speed, $0.18 per million tokens pricing, and 1048576-token context, with strengths in large-context handling, coding, and STEM performance.
It places third with a 13.5 intelligence_index, 102.47 t/s speed, $0.3 per million tokens pricing, and 10000000-token context, offering native multimodal support and strong reasoning over long sequences.
It ranks fourth via its 18.4 intelligence_index, 95.94 t/s speed, $0.6 per million tokens cost, and 1048576-token context, providing native multimodal text-and-image support plus strong general reasoning.
Handles complex reasoning across one million tokens of context.
Qwen3.6 Plus handles long multimodal sequences across text, images, and video.
Open-weight LLM with a 1M-token context for long text tasks.
It closes the list via its 1000000-token context, $0.97 per million tokens pricing, and optimization for fast coding assistance, making it efficient for developer workflows on large code contexts.
Qwen3.7 Max processes up to one million tokens in a single pass.
Fast open-weight multimodal model for million-token text, image, and video tasks.
It earns fifth position with a 1000000-token context, $1.56 per million tokens pricing, and support for text, image, and video inputs, enabling strong multimodal fusion for complex long-form tasks.
It ranks sixth through its 1000000-token context, $3.25 per million tokens pricing, and coding specialization, delivering clear structured outputs for large codebases and real-world programming tasks.
How we ranked this list
Ranked by real engagement (saves, reviews, usage and recency). Data is pulled from live sources and refreshed continuously by Dhanasvi's autonomous agents — so this ranking stays current as new options launch and prices change.
Frequently asked questions
Llama 4 Scout supports a 10000000-token context window.