Inception: mercury
5 benchmarks
Leaderboard position
#722
Overall rank (of 1126 models)
624
Elo rating
5
Benchmarks
Top 64.1%
Percentile
In Inception (of 4 models)
Above: ByteDance Seed: seedream-5.0-lite #721
Below: Houssem Fatnasssi: kilma-v1-base #723
Benchmark highlights
Best benchmark
Text Arena (Snapshot from 20 june 2026)
1306
Worst benchmark
BigCodeBench
45.5
Nearest competitor from another company
The model from another company with the most similar benchmark scores.
appeared as nearest in 2 of 5 benchmarks
2/5
40.0%
| Model | Votes | % of benchmarks |
|---|---|---|
| OpenAI: gpt-4o | 2 | 40.0% |
| Apple: diffucoder-7b-instruct | 1 | 20.0% |
| essentialai: rnj-1 | 1 | 20.0% |
| Google: Gemma 3 4B Instruct | 1 | 20.0% |
Benchmarks
Predicted scores
Estimated from 25 similar models and the global Elo rating. These are projections, not measured results.
| Benchmark | Predicted | Confidence | Based on |
|---|---|---|---|
| MMLU | ~64.3 | 57% | OpenAI: gpt-4o |
| GPQA | ~44.29 | 57% | OpenAI: gpt-4o |
| Artificial Analysis Intelligence Index v4.1 (Maximum Reasoning) | ~12.43 | 57% | OpenAI: gpt-4o |
| Artificial Analysis Intelligence Index v4.0 (Maximum Reasoning) | ~19.31 | 57% | OpenAI: gpt-4o |
| LLMArena (llmarena.ru) (Snapshot from 19 june 2026) | ~977.28 | 57% | OpenAI: gpt-4o |
| HumanEval+ | ~72.86 | 52% | Apple: diffucoder-7b-base, Apple: diffucoder-7b-instruct |
| MBPP+ | ~68.5 | 52% | Apple: diffucoder-7b-base, Apple: diffucoder-7b-instruct |
| ARC-Challenge | ~64.43 | 50% | — |
| ARC-Easy | ~79.82 | 50% | — |
| HellaSwag | ~75.6 | 50% | — |
| PIQA | ~83.77 | 50% | — |
| WinoGrande | ~71.9 | 50% | — |
| Humanity's Last Exam (no tools) | ~7.35 | 50% | — |
| Artificial Analysis Intelligence Index v5.0.1 | ~12 | 50% | — |
| MATH-500 | ~82.64 | 50% | — |
| AIME 2025 | ~58.77 | 50% | — |
| AIME 2026 | ~68.71 | 50% | — |
| HMMT Feb 2026 | ~60.24 | 50% | — |
| GPQA Diamond | ~52.93 | 50% | — |
| MMLU-Pro | ~56.29 | 50% | — |
| IFEval | ~79.38 | 50% | — |
| IFBench | ~49.55 | 50% | — |
| Terminal-Bench v2.1 | ~12.97 | 50% | — |
| SWE-bench Verified | ~32.4 | 50% | — |
| SWE-bench Pro | ~11.6 | 50% | — |