Ranking
Models whose context window is at least 1,000,000 tokens (from OpenRouter's catalogue), ordered by verification score. We don't test long-context recall itself. Top 15 of 37.
| # | Model | Score | Context | Cost / question | Speed (tok/s) |
|---|---|---|---|---|---|
| 1 |  Claude Opus 5.5Anthropic | 85.0% | 1M | $0.015 | 129 |
| 2 |  Gemini 4 ArgonGoogle | 84.8% | 1M | $0.0063 | 69 |
| 3 | Gemini 3.8 FlashGoogle | 82.5% | 1M | $0.0037 | 142 |
| 4 | Claude Fable 5.1Anthropic | 81.2% | 1M | $0.016 | 70 |
| 5 |  Muse WonderMeta | 78.8% | 1M | $0.021 | 175 |
| 6 | Claude Sonnet 5.5Anthropic | 78.8% | 1M | $0.0054 | 126 |
| 7 | Gemini 3.1 Pro PreviewGoogle | 77.5% | 1M | $0.017 | 124 |
| 8 | Muse Spark OpenMeta | 77.5% | 1M | $0.0045 | 220 |
| 9 |  Kimi K3.1 ProMoonshotAI | 77.5% | 1M | $0.01 | 38 |
| 10 |  GPT-6 AstraOpenAI | 76.2% | 1.1M | $0.013 | 49 |
| 11 | GPT-6.1 Terra ProOpenAI | 76.2% | 1.1M | $0.005 | — |
| 12 | GPT-6.1 SolOpenAI | 76.2% | 1.1M | $0.0029 | 41 |
| 13 | GPT-5.6 TerraOpenAI | 75.0% | 1.1M | $0.0063 | 101 |
| 14 | Gemini 3.5 Flash LiteGoogle | 75.0% | 1M | $0.0034 | 409 |
| 15 |  Qwen3.8 Max (0902)Qwen | 75.0% | 1M | $0.011 | 46 |