Which AI models are the most consistent over time? This report analyzes rank changes, state classifications, and sparkline volatility across 300 tracked models to produce a stability score from 0 to 100.
Rock Solid
273
Consistent
24
Variable
1
Volatile
2
Top 20 models with the highest stability scores. These models maintain consistent rankings with minimal volatility.
| # | Model | Score | Stability | 24h | 7d |
|---|---|---|---|---|---|
| 1 | Claude Fable 5Anthropic | 96.6 | 100 | 0 | 0 |
| 2 | Claude Opus 4.7 (Fast)Anthropic | 94.7 | 100 | 0 | 0 |
| 3 | Claude Opus 4.8 (Fast)Anthropic | 94.2 | 100 | 0 | 0 |
| 4 | Claude Opus 4.8Anthropic | 94.2 | 100 | 0 | 0 |
| 5 | GPT-5.5OpenAI | 92.2 | 100 | 0 | 0 |
| 6 | GPT-5.5 ProOpenAI | 90.3 | 100 | 0 | 0 |
| 7 | Claude Opus 4.6 (Fast)Anthropic | 90.0 | 100 | 0 | 0 |
| 8 | Grok 4.20xAI | 88.3 | 100 | 0 | 0 |
| 9 | Grok 4.20 Multi-AgentxAI | 87.4 | 100 | 0 | 0 |
| 10 | DeepSeek V4 ProDeepSeek | 86.2 | 100 | 0 | 0 |
| 11 | Claude Sonnet 4.6Anthropic | 84.7 | 100 | 0 | 0 |
| 12 | Grok 4.3xAI | 80.5 | 100 | 0 | 0 |
| 13 | Gemma 4 31B (free)Google | 80.0 | 100 | 0 | 0 |
| 14 | GPT-5.4 NanoOpenAI | 78.8 | 100 | 0 | 0 |
| 15 | GPT-5.4 MiniOpenAI | 78.8 | 100 | 0 | 0 |
| 16 | Gemini 3.5 FlashGoogle | 78.5 | 100 | 0 | 0 |
| 17 | GLM 5.2Zhipu AI | 78.1 | 100 | +1 | 0 |
| 18 | DeepSeek V4 FlashDeepSeek | 77.2 | 100 | +1 | 0 |
| 19 | GLM 5.1Zhipu AI | 76.1 | 100 | +1 | 0 |
| 20 | Kimi K2.6Moonshot AI | 75.2 | 100 | +1 | 0 |
Bottom 20 models with the lowest stability scores. These models show significant ranking fluctuations or inconsistent states.
| # | Model | Score | Stability | 24h | 7d |
|---|---|---|---|---|---|
| 1 | Mistral NemoMistral AI | 39.9 | 23 | -11 | -11 |
| 2 | GLM 5V TurboZhipu AI | 40.0 | 34 | -146 | +115 |
| 3 | Fugu Ultrasakana | 40.0 | 54 | +147 | +147 |
| 4 | Trinity Large Thinkingarcee-ai | 62.7 | 73 | +1 | -4 |
| 5 | Command R+ (08-2024)Cohere | 48.3 | 74 | +2 | +2 |
| 6 | Coder Largearcee-ai | 39.3 | 82 | -1 | -1 |
| 7 | Qwen3.5 Plus 2026-02-15Alibaba | 40.0 | 82 | -1 | -1 |
| 8 | Seed-2.0-MiniByteDance | 40.0 | 82 | -1 | -1 |
| 9 | Command R (08-2024)Cohere | 48.3 | 82 | +1 | +1 |
| 10 | Command ACohere | 50.4 | 82 | +1 | +1 |
| 11 | Claude 3 HaikuAnthropic | 50.9 | 82 | +1 | +1 |
| 12 | Kimi K2 0711Moonshot AI | 51.0 | 82 | +1 | +1 |
| 13 | Qwen3 235B A22BAlibaba | 53.5 | 82 | +1 | +1 |
| 14 | Llama 4 ScoutMeta | 54.9 | 82 | +1 | +1 |
| 15 | Mistral Large 2407Mistral AI | 55.8 | 82 | +1 | +1 |
| 16 | GPT-4o-mini (2024-07-18)OpenAI | 56.1 | 82 | +1 | +1 |
| 17 | gpt-oss-20b (free)OpenAI | 57.1 | 82 | +1 | +1 |
| 18 | Mixtral 8x22B InstructMistral AI | 63.0 | 82 | +1 | +1 |
| 19 | o3 Mini HighOpenAI | 63.5 | 82 | +1 | +1 |
| 20 | Llama 3.1 8B InstructMeta | 44.1 | 82 | +1 | +1 |
Aggregated stability metrics per provider. Providers are ranked by their average stability score across all models.
| Provider | Models | Avg Stability |
|---|---|---|
| xAI | 4 | 100.0 |
| Tencent | 2 | 100.0 |
| ~anthropic | 4 | 100.0 |
| perceptron | 1 | 100.0 |
| inclusionai | 3 | 100.0 |
| poolside | 4 | 100.0 |
| ~openai | 2 | 100.0 |
| 2 | 100.0 | |
| ~moonshotai | 1 | 100.0 |
| deepcogito | 1 | 100.0 |
| AI21 Labs | 1 | 100.0 |
| HUMAIN | 3 | 100.0 |
| TII | 6 | 100.0 |
| Baidu | 1 | 98.6 |
| Kuaishou | 1 | 97.7 |
| Perplexity | 5 | 96.6 |
| Amazon | 5 | 96.5 |
| NVIDIA | 11 | 96.4 |
| rekaai | 2 | 96.2 |
| Writer | 1 | 96.1 |
| Inception | 1 | 95.6 |
| Upstage | 1 | 94.1 |
| Anthropic | 15 | 94.1 |
| Alibaba | 48 | 93.7 |
| 22 | 93.6 | |
| Windsurf | 1 | 93.5 |
| Moonshot AI | 6 | 93.0 |
| StepFun | 2 | 92.8 |
| Microsoft | 2 | 92.5 |
| aion-labs | 3 | 92.4 |
| OpenAI | 58 | 89.8 |
| Liquid AI | 3 | 89.7 |
| MiniMax | 8 | 89.7 |
| IBM | 2 | 88.9 |
| Mistral AI | 18 | 88.6 |
| DeepSeek | 11 | 88.5 |
| Meta | 8 | 86.9 |
| ByteDance | 5 | 86.6 |
| arcee-ai | 4 | 86.3 |
| Zhipu AI | 12 | 85.9 |
| Xiaomi | 2 | 85.0 |
| Cursor | 2 | 85.0 |
| Allen AI | 1 | 84.7 |
| Cohere | 4 | 84.5 |
| sakana | 1 | 54.0 |
How stability scores are distributed across all 300 tracked models.
Our stability scoring system uses three key signals to measure how consistently a model performs over time.
The most direct measure of stability. Models lose up to 25 points for large 24-hour rank changes (5 points per rank position moved) and up to 21 points for 7-day changes (3 points per position). Models that hold their rank tightly score higher.
Each model has a state reflecting its overall reliability. Models in a "stable" state receive a 10-point bonus, while "fragile" models are penalized 15 points. This captures systemic reliability beyond simple rank movement.
The 14-day sparkline data reveals hidden volatility. We compute the standard deviation of the sparkline and subtract up to 20 points. Even models that end where they started can be penalized if they oscillated wildly along the way.
The stability score starts at 100 and is reduced based on three factors: 24-hour rank changes (up to -25 points, at 5 per position moved), 7-day rank changes (up to -21 points, at 3 per position), and sparkline volatility measured by standard deviation (up to -20 points). Models in a "stable" state get a +10 bonus, while "fragile" models lose 15 points.
Models are classified into four tiers based on their stability score: "Rock Solid" (85-100) means extremely consistent performance with minimal fluctuation. "Consistent" (70-84) means generally reliable with minor variations. "Variable" (50-69) shows noticeable ranking fluctuations. "Volatile" (below 50) indicates significant instability and unpredictable performance.
Stability indicates how predictably a model will perform over time. A highly rated but volatile model may deliver inconsistent results, which is problematic for production applications requiring reliable output quality. Stable models provide more predictable performance, making them safer choices for mission-critical workloads even if they do not always hold the top rank.