AI models compared

584 models with published results from our own benchmarks and licensed sources, strongest first: by our overall score, then the Artificial Analysis Intelligence Index. Open a model for every result, with its source and date. 225 of them are on sale through OpenRouter, with prices on the API pricing page; launches are on the release tracker.

Sorted by Spring Prompt overall, highest first; models without a value come last. Select a column heading to sort by it.

ModelSpring Prompt overallIntelligence IndexPrice, blendedReleasedContextSources
Qwen3-VL-30B-A3B-InstructAlibaba · open weights – 7.9 $0.26 3 Oct 2025 262k 1
GPT-4.1 nanoOpenAI – 7.8 $0.18 14 Apr 2025 1,047k 4
Gemma 4 E2BGoogle · not on sale – 7.8 – 2 Apr 2026 – 2
Qwen3-Omni-30B-A3B-ThinkingAlibaba · not on sale – 7.8 – 22 Sep 2025 – 2
Solar Pro 3Upstage – 7.8 $0.26 6 Apr 2026 131k 2
GPT-4o (2024-08-06)OpenAI – 7.7 $4.38 6 Aug 2024 128k 4
Llama 3.3 70B InstructMeta · open weights – 7.7 $0.64 6 Dec 2024 131k 4
Qwen2.5-72B-InstructAlibaba · open weights – 7.7 $0.37 17 Sep 2024 32k 3
Qwen3-30B-A3BAlibaba · open weights – 7.6 $0.21 28 Apr 2025 131k 3
GLM-4.5VZ.ai · open weights – 7.6 $0.90 11 Aug 2025 65k 2
Devstral Small 1.1Mistral AI · not on sale – 7.6 – 10 Jul 2025 – 1
Qwen3-30B-A3B-Instruct-2507Alibaba · open weights – 7.5 $0.14 29 Jul 2025 262k 4
Devstral Small 2Mistral AI · not on sale – 7.5 – 9 Dec 2025 – 2
Qwen3-32BAlibaba · open weights – 7.3 $0.21 28 Apr 2025 131k 5
GPT-4o (2024-05-13)OpenAI – 7.3 $7.50 13 May 2024 128k 3
Kimi Linear 48B A3B InstructMoonshot AI · not on sale – 7.3 – 30 Oct 2025 – 2
Qwen3-VL-8B-InstructAlibaba · open weights · not on sale – 7.3 $0.20 14 Oct 2025 262k 2
Mistral Small 3.1 24BMistral AI · open weights – 7.1 $0.40 17 Mar 2025 128k 3
Command ACohere · open weights – 7.0 $4.38 20 May 2026 256k 5
Mistral Small 3.2 24BMistral AI · open weights – 7.0 $0.13 20 Jun 2025 256k 4
Nova Pro 1.0Amazon – 7.0 $1.40 3 Dec 2024 300k 4
Qwen3-VL-4B-ThinkingAlibaba · not on sale – 7.0 – 14 Oct 2025 – 2
Llama 3.1 8B InstructMeta · open weights – 6.9 $0.06 23 Jul 2024 131k 4
Qwen3.5-2BAlibaba · not on sale – 6.9 – 2 Mar 2026 – 2
Mistral Large 2 (2407)Mistral AI – 6.8 $3.00 24 Jul 2024 131k 3
Mistral Small 3Mistral AI · open weights – 6.7 $0.06 30 Jan 2025 32k 4
Qwen3-14BAlibaba · open weights – 6.7 $0.15 28 Apr 2025 131k 4
GPT-4o mini (2024-07-18)OpenAI – 6.7 $0.26 18 Jul 2024 128k 3
Nova Lite 1.0Amazon – 6.7 $0.10 3 Dec 2024 300k 3
Qwen3-4B-Instruct-2507Alibaba · not on sale – 6.7 – 6 Aug 2025 – 3
Qwen2.5-Coder-32B-InstructAlibaba · open weights – 6.7 $0.74 18 Sep 2024 32k 2
Llama 3.1 70B InstructMeta · open weights – 6.6 $0.40 23 Jul 2024 131k 3
Llama 4 ScoutMeta · open weights – 6.5 $0.28 5 Apr 2025 1,310k 5
Qwen3.5-0.8BAlibaba · not on sale – 6.1 – 2 Mar 2026 – 2
Qwen3-8BAlibaba · open weights · not on sale – 6.0 $0.20 28 Apr 2025 131k 4
Ministral 3 14BMistral AI · open weights – 6.0 $0.20 2 Dec 2025 262k 3
Qwen3-Omni-30B-A3B-InstructAlibaba · not on sale – 6.0 – 22 Sep 2025 – 1
Phi-4Microsoft · open weights – 5.9 $0.09 12 Dec 2024 16k 5
Nova Micro 1.0Amazon – 5.9 $0.06 3 Dec 2024 128k 4
Llama 3.2 3B InstructMeta · open weights – 5.7 $0.12 25 Sep 2024 131k 4
Mixtral 8x22B InstructMistral AI · open weights – 5.7 $3.00 17 Apr 2024 65k 3
Qwen3-VL-4B-InstructAlibaba · not on sale – 5.7 – 14 Oct 2025 – 2
Reka Flash 3Reka AI · open weights – 5.6 $0.12 10 Mar 2025 65k 2
Ministral 3 8BMistral AI · open weights – 5.5 $0.15 2 Dec 2025 262k 3
Gemma 3 270MGoogle · not on sale – 5.1 – 11 Mar 2025 – 1
Gemma 3 27BGoogle · open weights – 4.9 $0.14 11 Mar 2025 131k 5
Gemma 3 4BGoogle · open weights – 4.8 $0.06 11 Mar 2025 131k 5
Llama 3.2 1B InstructMeta · open weights – 4.8 $0.07 25 Sep 2024 60k 4
Ministral 3 3BMistral AI · open weights – 4.8 $0.10 2 Dec 2025 131k 2
Gemma 3 12BGoogle · open weights – 3.8 $0.08 11 Mar 2025 131k 5

Spring Prompt overall is our score across every source, out of 100 (how it works). The Intelligence Index is Artificial Analysis's, at each model's best setting. Price is the developer's list price, or the typical OpenRouter host, for three input tokens to one output, per million tokens.

Want two models side by side? Compare models.