6 models
Ranked by SWE-Bench Pro
Reasoning 61.5

Muse Spark 1.1 is a multimodal reasoning model from Meta, built for agentic tasks. It accepts text, images, video, audio, and PDF documents and returns text, with a 1M-token context...

meta/muse-spark-1.1 2026-04-08 1.04858M context $1.25/M input $4.25/M output
12 providers
Reasoning Tools 60.6

Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks

alibaba/qwen3.7-max 2026-05-21 1M context $2.5/M input $7.5/M output
31 providers
Reasoning Tools JSON Open weights 56.3

Step 3.7 Flash is StepFun's latest high-efficiency multimodal Mixture-of-Experts model. It pairs a 196B-parameter language backbone with a vision encoder for native image and video understanding, activating roughly 11B parameters...

stepfun/step-3.7-flash 2026-05-29 256K context $0.2/M input $1.15/M output
17 providers
Reasoning Tools JSON Open weights 51.2

Muse Glimmer 30B is a dense, open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark and optimized for autonomous agents on consumer hardware. It is suited for long-horizon...

meta/muse-glimmer-30b 2026-08-10 131.072K context $0.3/M input $1.1/M output
9 providers
Reasoning Tools JSON Open weights 27.3

Kimi K2.6 is Moonshot AI's next-generation multimodal model, designed for long-horizon coding, coding-driven UI/UX generation, and multi-agent orchestration. It handles complex end-to-end coding tasks across Python, Rust, and Go, and...

moonshotai/kimi-k2.6 2026-04-21 262.144K context $0.95/M input $4/M output
64 providers
Reasoning Tools JSON Open weights 18.0

DeepSeek V4 Pro is a large-scale Mixture-of-Experts model from DeepSeek with 1.6T total parameters and 49B activated parameters, supporting a 1M-token context window. It is designed for advanced reasoning, coding,...

deepseek/deepseek-v4-pro 2026-04-24 1.024M context $0.84/M input $1.679/M output
62 providers