OpenAI MRCR v2 8-needle 64K-128K (MRCR v2 64K-128K)
We show this table for reference; we do not rank on it.
MRCR v2 slice focused on long-context retrieval at 64K-128K lengths.
Benchmark score on MRCR v2 64K-128K — November 13, 2026
We compile the MRCR v2 64K-128K rows from provider self-reports. Gemini 3.7 Flash leads the table at 97%, followed by GPT-5.5 (83.1%). We do not use these results to rank models overall.
Gemini 3.7 Flash
GPT-5.5
OpenAI
Reported by OpenAI2 modelsReasoningCurrentDisplay onlyUpdated November 13, 2026
Benchmark score table (2 models)
All 2 results on this page were reported by the labs that built the models.
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Gemini 3.7 FlashGoogleReported by Google | 97% | Reported by Google | Not reported | Closed |
| 2 | GPT-5.5OpenAIReported by OpenAI | 83.1% | Reported by OpenAI | Not reported | Closed |
- Lab-reported
- Benchmark-run
- Independent
About MRCR v2 64K-128K
Year
2026
Tasks
8-needle retrieval tasks
Format
Long-context retrieval
Difficulty
Long-context reasoning
Measures whether models can recover the right details when multiple relevant items are buried in long contexts.
Freshness and provenance
Version
MRCR v2 64K-128K 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does MRCR v2 64K-128K measure?
MRCR v2 slice focused on long-context retrieval at 64K-128K lengths.
Which model scores highest on MRCR v2 64K-128K?
Gemini 3.7 Flash by Google currently leads with a score of 97% on MRCR v2 64K-128K.
How many models are evaluated on MRCR v2 64K-128K?
2 AI models have published results on MRCR v2 64K-128K in the BenchLM catalog.
Compare top models on MRCR v2 64K-128K
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.