Analytics
How the field moves, built only from measured results and dated releases. See methodology.
Best Intelligence Index over time
Running best score among models released so far, open-weight versus proprietary. 57 models with a measured index and a trustworthy release date; Artificial Analysis mostly indexes models released from mid-2026, so earlier history is thin.
Show the data as a table
| Release date | Model | Lab | Open weights | Index |
|---|---|---|---|---|
| 2025-08-05 | gpt-oss-120b | OpenAI | Yes | 3.8 |
| 2026-03-16 | Mistral Small 4 | Mistral AI | Yes | 4.3 |
| 2026-03-16 | NVIDIA Nemotron 3 Nano 4B | NVIDIA AI & NIM | Yes | 1.8 |
| 2026-04-02 | Gemma 4 31B | Google DeepMind | Yes | 4.4 |
| 2026-04-29 | Mistral Medium 3.5 | Mistral AI | Yes | 4.6 |
| 2026-06-01 | MiniMax M3 | MiniMax | Yes | 18.9 |
| 2026-06-04 | Nemotron 3 Ultra | NVIDIA AI & NIM | Yes | 11.8 |
| 2026-06-09 | North Mini Code | Cohere | Yes | 3.4 |
| 2026-07-15 | Inkling | Thinking Machines Lab | Yes | 12.5 |
| 2026-07-21 | Gemini 3.5 Flash-Lite | Google DeepMind | No | 11.7 |
| 2026-07-30 | Inkling Small | Thinking Machines Lab | Yes | 12.9 |
| 2026-08-06 | Solar Pro 4 | Upstage | No | 14.2 |
| 2026-08-10 | Muse Glimmer | Meta AI | Yes | 6.3 |
| 2026-08-11 | Nemotron 3.5 Lightning | NVIDIA AI & NIM | Yes | 5.5 |
| 2026-08-12 | K-EXAONE 2.0 | LG AI Research | Yes | 7.6 |
| 2026-08-13 | DeepSeek V4-Pro | DeepSeek | Yes | 23.1 |
| 2026-08-14 | Qwen3.8 27B | Alibaba Cloud / Qwen | Yes | 20.1 |
| 2026-08-18 | GLM 5.3 | Zhipu AI (Z.ai) | Yes | 31.7 |
| 2026-08-25 | Granite 4.2 8B | IBM | Yes | 5.1 |
| 2026-08-25 | Granite 4.2 3B | IBM | Yes | 4.7 |
| 2026-08-26 | GLM 5.3 Flash | Zhipu AI (Z.ai) | Yes | 27.8 |
| 2026-08-26 | Qwen3.8-Flash-Next | Alibaba Cloud / Qwen | Yes | 28.6 |
| 2026-08-30 | Apodex 1.1 | Apodex | No | 13.7 |
| 2026-09-02 | Qwen3.8 Max (0902) | Alibaba Cloud / Qwen | Yes | 33.7 |
| 2026-09-02 | Gemini 3.8 Flash | Google DeepMind | No | 30.8 |
| 2026-09-03 | GPT-6 Astra | OpenAI | No | 56.9 |
| 2026-09-03 | K2 Horizon 375B A23B | Institute of Foundation Models (MBZUAI) | Yes | 17.8 |
| 2026-09-18 | Step 5 | StepFun | Yes | 28.5 |
| 2026-09-21 | Grok 4.7 | xAI | No | 35.5 |
| 2026-09-21 | MiMo-V2.6-Pro | Xiaomi | Yes | 29.9 |
| 2026-09-21 | MiMo-V2.6-Flash | Xiaomi | Yes | 26.7 |
| 2026-09-22 | Claude Opus 5.5 | Anthropic | No | 59.1 |
| 2026-09-22 | GPT-6 Luna | OpenAI | No | 27.9 |
| 2026-09-28 | Claude Sonnet 5.5 | Anthropic | No | 54.5 |
| 2026-09-29 | GPT-6.1 Sol | OpenAI | No | 54.6 |
| 2026-09-30 | Gemini 4 Argon | Google DeepMind | No | 50.4 |
| 2026-10-07 | Claude Haiku 5.5 | Anthropic | No | 36.5 |
| 2026-10-12 | Muse Wonder Full | Meta AI | No | 50.1 |
| 2026-10-12 | Muse Wonder | Meta AI | No | 43.6 |
| 2026-10-15 | Nemotron 3.5 Super VL | NVIDIA AI & NIM | Yes | 13.6 |
| 2026-10-15 | Muse Spark Open | Meta AI | Yes | 24.9 |
| 2026-10-16 | Beam 1.0 | Reflection AI | Yes | 17.7 |
| 2026-10-17 | DeepSeek V4.1 Flash 1017 | DeepSeek | Yes | 32.6 |
| 2026-10-19 | MiniMax M3.1 Flash | MiniMax | Yes | 29.1 |
| 2026-10-22 | Ling 3.1 Flash 1019 | inclusionAI (Ant Group) | Yes | 29.6 |
| 2026-10-23 | Kimi K3.1 Pro | Moonshot AI | Yes | 46.5 |
| 2026-10-25 | Nanbeige4.3-33B | Nanbeige | Yes | 20.4 |
| 2026-10-26 | GPT-6.1 Terra | OpenAI | No | 51.3 |
| 2026-10-28 | Gemini 4 Carbon | Google DeepMind | No | 60.0 |
| 2026-10-29 | Mistral Large 4 | Mistral AI | Yes | 26.1 |
| 2026-11-03 | Hy4 | Tencent | No | 34.6 |
| 2026-11-05 | Inkling 2 Small | Thinking Machines Lab | Yes | 30.6 |
| 2026-11-07 | Agnes 2.5 Pro Theta | Sapiens AI | No | 24.0 |
| 2026-11-09 | Claude Fable 5.2 | Anthropic | No | 64.6 |
| 2026-11-09 | Muse Ember | Meta AI | Yes | 36.0 |
| 2026-11-11 | JT-4.1 Flash 236B A21B | China Mobile | No | 19.0 |
| 2026-11-12 | Gemini 4 Hydrogen | Google DeepMind | No | 53.9 |
Price against performance
Intelligence Index against what it cost Artificial Analysis to run the whole Index, for 48 models. Up and to the left is better. Cheapest model within 15% of the top score: Gemini 4 Carbon at $5.48.
Show the data as a table
| Model | Lab | Open weights | Index | Cost to run Index (USD) |
|---|---|---|---|---|
| Claude Fable 5.2 | Anthropic | No | 64.6 | 26.54 |
| Gemini 4 Carbon | Google DeepMind | No | 60.0 | 5.48 |
| Claude Opus 5.5 | Anthropic | No | 59.1 | 10.04 |
| GPT-6 Astra | OpenAI | No | 56.9 | 6.24 |
| GPT-6.1 Sol | OpenAI | No | 54.6 | 1.27 |
| Claude Sonnet 5.5 | Anthropic | No | 54.5 | 10.02 |
| Gemini 4 Hydrogen | Google DeepMind | No | 53.9 | 3.16 |
| GPT-6.1 Terra | OpenAI | No | 51.3 | 0.96 |
| Gemini 4 Argon | Google DeepMind | No | 50.4 | 4.97 |
| Kimi K3.1 Pro | Moonshot AI | Yes | 46.5 | 3.68 |
| Muse Wonder | Meta AI | No | 43.6 | 6.95 |
| Claude Haiku 5.5 | Anthropic | No | 36.5 | 0.41 |
| Muse Ember | Meta AI | Yes | 36.0 | 0.14 |
| Grok 4.7 | xAI | No | 35.5 | 7.77 |
| Hy4 | Tencent | No | 34.6 | 1.03 |
| Qwen3.8 Max (0902) | Alibaba Cloud / Qwen | Yes | 33.7 | 10.02 |
| DeepSeek V4.1 Flash 1017 | DeepSeek | Yes | 32.6 | 0.60 |
| GLM 5.3 | Zhipu AI (Z.ai) | Yes | 31.7 | 5.01 |
| Gemini 3.8 Flash | Google DeepMind | No | 30.8 | 7.77 |
| Inkling 2 Small | Thinking Machines Lab | Yes | 30.6 | 1.18 |
| MiMo-V2.6-Pro | Xiaomi | Yes | 29.9 | 0.26 |
| Ling 3.1 Flash 1019 | inclusionAI (Ant Group) | Yes | 29.6 | 3.19 |
| MiniMax M3.1 Flash | MiniMax | Yes | 29.1 | 0.49 |
| Qwen3.8-Flash-Next | Alibaba Cloud / Qwen | Yes | 28.6 | 0.57 |
| Step 5 | StepFun | Yes | 28.5 | 2.35 |
| GPT-6 Luna | OpenAI | No | 27.9 | 0.15 |
| GLM 5.3 Flash | Zhipu AI (Z.ai) | Yes | 27.8 | 0.58 |
| MiMo-V2.6-Flash | Xiaomi | Yes | 26.7 | 0.13 |
| Mistral Large 4 | Mistral AI | Yes | 26.1 | 2.06 |
| Muse Spark Open | Meta AI | Yes | 24.9 | 2.02 |
| Agnes 2.5 Pro Theta | Sapiens AI | No | 24.0 | 0.57 |
| DeepSeek V4-Pro | DeepSeek | Yes | 23.1 | 1.77 |
| Qwen3.8 27B | Alibaba Cloud / Qwen | Yes | 20.1 | 3.53 |
| MiniMax M3 | MiniMax | Yes | 18.9 | 1.29 |
| Beam 1.0 | Reflection AI | Yes | 17.7 | 0.28 |
| Solar Pro 4 | Upstage | No | 14.2 | 0.96 |
| Apodex 1.1 | Apodex | No | 13.7 | 0.76 |
| Nemotron 3.5 Super VL | NVIDIA AI & NIM | Yes | 13.6 | 0.56 |
| Inkling Small | Thinking Machines Lab | Yes | 12.9 | 0.20 |
| Inkling | Thinking Machines Lab | Yes | 12.5 | 1.35 |
| Nemotron 3 Ultra | NVIDIA AI & NIM | Yes | 11.8 | 2.21 |
| Gemini 3.5 Flash-Lite | Google DeepMind | No | 11.7 | 0.69 |
| Muse Glimmer | Meta AI | Yes | 6.3 | 0.18 |
| Nemotron 3.5 Lightning | NVIDIA AI & NIM | Yes | 5.5 | 0.31 |
| Granite 4.2 8B | IBM | Yes | 5.1 | 0.04 |
| Mistral Medium 3.5 | Mistral AI | Yes | 4.6 | 1.04 |
| Mistral Small 4 | Mistral AI | Yes | 4.3 | 0.04 |
| gpt-oss-120b | OpenAI | Yes | 3.8 | 0.35 |
Model releases per quarter
Releases with a verified, Artificial Analysis or version-id date, over the last 8 quarters in the data (351 models). Eight labs have a fixed colour; everything else is grouped. Models whose only date is an OpenRouter listing are not counted.
Show the data as a table
| Quarter | Google DeepMind | Anthropic | DeepSeek | Alibaba Cloud / Qwen | Meta AI | OpenAI | Moonshot AI | xAI | Other labs |
|---|---|---|---|---|---|---|---|---|---|
| 2025 Q1 | 6 | 1 | 1 | 3 | 0 | 2 | 1 | 1 | 19 |
| 2025 Q2 | 2 | 0 | 0 | 1 | 2 | 1 | 0 | 0 | 3 |
| 2025 Q3 | 1 | 2 | 0 | 4 | 0 | 2 | 0 | 1 | 15 |
| 2025 Q4 | 3 | 2 | 1 | 0 | 0 | 7 | 2 | 0 | 31 |
| 2026 Q1 | 8 | 2 | 0 | 13 | 0 | 6 | 0 | 2 | 27 |
| 2026 Q2 | 11 | 4 | 3 | 5 | 0 | 5 | 2 | 2 | 26 |
| 2026 Q3 | 6 | 4 | 6 | 9 | 7 | 14 | 1 | 3 | 40 |
| 2026 Q4 | 4 | 2 | 1 | 0 | 4 | 2 | 1 | 0 | 17 |
Source: Verified announcements, Artificial Analysis release tracker, OpenRouter version ids