Mercury 2 benchmarks & scores
Independent Mercury 2 AI coding benchmark results Intelligence 21, Coding 28, Agentic 10, with a Vibe Coding Index of 20.4. Compare its overall rank, token price, context window, and practical agentic workflow strengths below.
Evidence confidence
High confidence
Complete core scores supported by at least two current verified sources.
Catalog capabilities
What the model supports
Suite radar
Suite scores
Task placements
Where Mercury 2 places outside the VCI blend
Head-to-head preference boards that never rewrite the suite scores above — Arena Code WebDev and Design Arena categories only.
Arena Code WebDev
Human preference for web development outputs. Separate from Artificial Analysis coding until blended into the published Coding suite score.
Coding suite score above (28.4, multi-source index) includes this signal when current.
Design Arena categories
Task leaderboards for Mercury 2
Website, UI, game, data viz, and other tournament boards — not used in the Vibe Coding Index formula.
| Category | Rank | ELO | Win rate | Avg time | Tournaments | |
|---|---|---|---|---|---|---|
|
ASCII art
|
#65 / 66 | 1,025 | 27.0% | 10.4s | 4,036 | Source ↗ |
|
SVG
|
#84 / 90 | 1,026 | 25.4% | 2.4s | 4,036 | Source ↗ |
|
3D
|
#110 / 121 | 1,049 | 24.6% | 3.4s | 4,036 | Full board → |
|
Game development
|
#111 / 129 | 1,036 | 21.3% | 3.8s | 4,036 | Full board → |
|
UI components
|
#111 / 126 | 1,016 | 20.2% | 3.9s | 4,036 | Full board → |
|
Data visualization
|
#114 / 128 | 1,009 | 20.9% | 3.1s | 4,036 | Full board → |
|
Code categories
|
#118 / 132 | 1,028 | 21.3% | 4.3s | 4,036 | Full board → |
|
Web development
|
#126 / 141 | 1,016 | 20.1% | 5.1s | 4,036 | Full board → |
Source record
Source. Artificial Analysis (artificialanalysis.ai) via OpenRouter (openrouter.ai/rankings).
TheVibeFather rankings
Mercury 2 practical agentic coding scorecard
This is our algorithmic assessment of where Mercury 2 ranks for common vibe coding and agentic engineering workflows. Every stable 0–10 score is calculated consistently from the sourced Intelligence, Coding, and Agentic signals above—never from an unsourced opinion score.
Small, well-defined code changes
Precision on scoped edits, fixes, and implementation tasks.
UI and CSS iteration
Front-end implementation with iterative tool-driven refinement.
Routine debugging
Diagnosing failures and turning reasoning into correct code changes.
Repository-wide refactors
Coordinating larger edits across files while preserving intent.
Architecture decisions
Reasoning through tradeoffs, constraints, and system-level choices.
Tool use and workflow execution
Planning and completing multi-step work with tools and feedback loops.
Long autonomous coding tasks
Sustaining coherent progress across longer agentic engineering runs.
Overall VibeFather rating
Our complete Vibe Coding Index, expressed on the same 0–10 practical scale.
We start with Intelligence, Coding, Agentic benchmark evidence, and preserve missing values as unverified.
Our code-controlled algorithm blends the native 0–100 signals by workflow and expresses the result on a stable 0–10 scale. Every required input must be present.
Each workflow score is ranked against every model with comparable evidence, so positions update when the benchmark field changes.
The Vibe Coding Index is TheVibeFather’s code-controlled 0–100 composite for practical coding and agentic capability. Price and adoption are reported independently and never alter quality. Exact weighting remains proprietary, inputs, missing-data behavior, and per-category ranks are disclosed here. Read the benchmark methodology →
Recorded score history
How Mercury 2 has moved
No material score or rank change.
No material score or rank change.
No material score or rank change.
Where Mercury 2 sits
Its logo stays full-color and highlighted while the rest of the field recedes for quick comparison.
Alibaba
Amazon
Anthropic
Cohere
DeepSeek
Google
Inception
Inclusionai
Kwaipilot
Meta
Meta Llama
MiniMax
Mistral
Moonshot
NVIDIA
Nex Agi
OpenAI
Stepfun
Tencent
Thinkingmachines
Upstage
Xiaomi
Zhipu
xAI
Hover, tap or focus a model for axis guides · resting dashed lines = medians
27 model(s) hidden — data not yet verified Model benchmark FAQ
Mercury 2 for vibe coding and agentic engineering
Answers update from this model’s current sourced scores, practical rankings, and verified pricing.
What is Mercury 2's Vibe Coding Index?
Mercury 2 has a Vibe Coding Index of 20.4/100 and is #66 of 103 in the current field. The index is TheVibeFather's stable composite for practical AI coding, combining independently sourced Intelligence, Coding, and Agentic benchmark signals. Price and adoption are compared separately and never inflate the quality score.
How does Mercury 2 rank for vibe coding?
For vibe coding, Mercury 2's strongest practical area is Small, well-defined code changes at 2.6/10, ranking #67 of 103 models in TheVibeFather rankings.
Is Mercury 2 good for agentic coding and autonomous tasks?
Mercury 2 scores 1.4/10 for tool use and workflow execution and 1.2/10 for long autonomous coding tasks. Those categories emphasize the Agentic benchmark signal used for multi-step planning, tool calls, edits, tests, and feedback loops.
What coding benchmark scores does Mercury 2 have?
The current sourced benchmark profile for Mercury 2 is Intelligence 21/100, Coding 28/100, Agentic 10/100. TheVibeFather converts those 0–100 signals into consistent practical 0–10 workflow scores so models can be compared for real agentic engineering work.
How much does Mercury 2 cost for agentic engineering?
The current benchmark uses a 3 to 1 input-to-output blend of $0.38 per million blended tokens. That places it #34 of 108 by price, where a lower rank number means less expensive.
Are vibe coding, agentic coding, and agentic engineering the same?
They describe overlapping AI-assisted software workflows. Vibe coding emphasizes directing software through intent and iteration, agentic coding emphasizes a model planning and executing multi-step work with tools, agentic engineering is the broader discipline of designing, supervising, and validating those workflows. TheVibeFather benchmarks the shared capabilities behind all three terms on every model page, including Mercury 2.
