Benchmark
OpenAI MRCR v2 · 8-needle
Long context benchmarks measure how well models use information in long inputs. Scores are percentages. Higher is better.
- Models measured
- 28
- Scores recorded
- 74
- Sources
- 1
- IQ weight
- 1.76
Score distribution
- 78.1%
- Mean
- 84.9%
- Median
- 19.3%
- Std dev
- 44.1%
- 10th percentile
- 97.2%
- 90th percentile
- 26.3%
- Min
This benchmark has a weight of 1.76 for chat models, measured over 28 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | GPT-6 Astra256K-512K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 100% |
| 2 | Muse Spark 1.3max · 256K–512K · 100 examples · 8-needle · no tools · mean sequence-matcher ratio· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 98.5% |
| 3 | GPT 5.24K-8K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 98.2% |
| 4 | GPT 5.24K–8K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 98.2% |
| 5 | GPT 5.54K–8K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 98.1% |
| 6 | Muse Spark 1.3max · 512K–1M · 100 examples · 8-needle · no tools · mean sequence-matcher ratio· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 98.1% |
| 7 | GPT 5.44K-8K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 97.3% |
| 8 | GPT 5.416K-32K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 97.2% |
| 9 | Gemini 3.7 Flash128K cumulative average· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 97% |
| 10 | GPT 5.516K–32K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 96.5% |
| 11 | GPT-6 Astra512K-1M· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 96.3% |
| 12 | GPT 5.216K-32K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 95.3% |
| 13 | GPT 5.216K–32K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 95.3% |
| 14 | GPT 5.5Gemini 3.5 Flash Model Card · May 2026 · 128K cumulative average· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 94.8% |
| 15 | Claude Opus 4.6Anthropic Opus 4.6 launch page · adaptive thinking · max effort unless noted · 256K pointwise · 8-needle· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 93% |
| 16 | GPT 5.58K–16K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 93% |
| 17 | GPT 5.232K-64K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 92% |
| 18 | GPT 5.232K–64K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 92% |
| 19 | Gemini 3.6 FlashGemini 3.6 Flash Model Card · July 2026 · 128K cumulative average· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 91.8% |
| 20 | Qwen3.7 PlusQwen3.7-Plus release · 128K · 8 needles · Google DeepMind eval_hub protocol· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 91.7% |
| 21 | GPT-5.6 Sol256K-512K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 91.5% |
| 22 | GPT-5.6 Solmax · 256K–512K · 100 examples · 8-needle · no tools · mean sequence-matcher ratio· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 91.5% |
| 23 | GPT 5.48K-16K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 91.4% |
| 24 | Claude Sonnet 4.6256K bin · mean match ratio · 64K extended thinking · 5 trials· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 90.6% |
| 25 | GPT 5.432K-64K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 90.5% |
| 26 | Qwen3.7 MaxQwen3.7-Max release · 128K · 8 needles · Google DeepMind eval_hub protocol· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 90.4% |
| 27 | Claude Sonnet 4.6256K bin · mean match ratio · adaptive thinking · max effort · 5 trials· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 90.3% |
| 28 | GPT 5.532K–64K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 90% |
| 29 | GPT-5.6 Terra256K-512K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 89.6% |
| 30 | GPT 5.28K-16K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 89.3% |
| 31 | GPT 5.28K–16K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 89.3% |
| 32 | GPT 5.5128K–256K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 87.5% |
| 33 | GPT 5.464K-128K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 86% |
| 34 | GPT 5.264K-128K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 85.6% |
| 35 | GPT 5.264K–128K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 85.6% |
| 36 | Claude Sonnet 4.6Gemini 3.5 Flash Model Card · May 2026 · 128K cumulative average· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 84.9% |
| 37 | Gemini 3.1 Pro Preview128K cumulative average· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 84.9% |
| 38 | Gemini 3.1 Pro PreviewGemini 3.5 Flash Model Card · May 2026 · 128K cumulative average· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 84.9% |
| 39 | Gemini 3.1 Pro PreviewGemini 3.6 Flash Model Card · July 2026 · 128K cumulative average· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 84.9% |
| 40 | GPT 5.564K–128K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 83.1% |
| 41 | GPT 5.5256K-512K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 81.5% |
| 42 | GPT 5.5256K–512K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 81.5% |
| 43 | Grok 4.5Gemini 3.6 Flash Model Card · July 2026 · 128K cumulative average· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 81.4% |
| 44 | GPT 5.4128K-256K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79.3% |
| 45 | Gemini 3.5 FlashGemini 3.5 Flash Model Card · May 2026 · 128K cumulative average· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 77.3% |
| 46 | Gemini 3.5 FlashGemini 3.6 Flash Model Card · July 2026 · 128K cumulative average· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 77.3% |
| 47 | Gemini 3 Pro PreviewGemini 3 Pro Model Card · results as of November 2025 · 128K cumulative average· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 77% |
| 48 | GPT 5.2128K-256K· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 77% |
| 49 | GPT 5.2128K–256K · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 77% |
| 50 | GPT-5.6 LunaGemini 3.6 Flash Model Card · July 2026 · 128K cumulative average· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 74.8% |
Showing 1–50 of 74 results.