Scale Labs MCP Atlas (MCP Atlas)
We show this table for reference; we do not rank on it.
A Scale Labs public leaderboard mirrored as display-only reference data. It does not affect BenchLM rankings.
Scale score on MCP Atlas — November 12, 2026 snapshot
We mirror the published scale score view for MCP Atlas. Gemini 4 Argon leads the public snapshot at 90.7%, followed by Muse Spark 1.1 (88.1%) and Fable 5.1 (87.2%). We do not use these results to rank models overall.
Gemini 4 Argon
Muse Spark 1.1
Meta
Fable 5.1
Anthropic
37 modelsAgenticCurrentDisplay onlyUpdated November 12, 2026 snapshot
Scale score table (37 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Gemini 4 ArgonGoogle | 90.7% | Not reported | Closed |
| 2 | Muse Spark 1.1Meta | 88.1% | Not reported | Closed |
| 3 | Fable 5.1Anthropic | 87.2% | Not reported | Unknown |
| 4 | claude-opus-5 (xhigh)Anthropic | 85.8% | Not reported | Unknown |
| 5 | Opus 5.5Anthropic | 85.4% | Not reported | Unknown |
| 6 | Inkling 2 SmallThinking Machines Lab | 85.4% | Not reported | Closed |
| 7 | Qwen3.8-2.4T-A95B (xHigh)Alibaba | 84.5% | Not reported | Unknown |
| 8 | GLM-5.3Z.AI | 84.2% | Not reported | Open |
| 9 | Gemini 3.5 Flash (high)Google | 83.6% | Not reported | Unknown |
| 10 | Claude Fable 5Anthropic | 83.3% | Not reported | Closed |
| 11 | kimi-k3 (max)Kimi | 82.3% | Not reported | Unknown |
| 12 | Muse SparkMeta | 82.2% | Not reported | Closed |
| 13 | claude-opus-4-8 (max)Anthropic | 82.2% | Not reported | Unknown |
| 14 | gpt-5.6 (sol)OpenAI | 81.8% | Not reported | Unknown |
| 15 | Inkling-SmallThinking Machines Lab | 79.2% | Not reported | Open |
| 16 | claude-opus-4-7 (max)Anthropic | 79.1% | Not reported | Unknown |
| 17 | gemini-3.1-pro-preview (high)Google | 78.2% | Not reported | Unknown |
| 18 | glm-5p2Z.AI | 77.8% | Not reported | Unknown |
| 19 | claude-opus-4-6 (max)Anthropic | 76.8% | Not reported | Unknown |
| 20 | Inkling (xHigh)Thinkingmachines | 76% | Not reported | Unknown |
| 21 | glm-5p1Z.AI | 75.6% | Not reported | Unknown |
| 22 | gpt-5.5 (xhigh)OpenAI | 75.3% | Not reported | Unknown |
| 23 | gpt-5.4 (xhigh)OpenAI | 70.6% | Not reported | Unknown |
| 24 | gemini-3-pro-previewGoogle | 70.3% | Not reported | Unknown |
| 25 | claude-opus-4-5 (high)Anthropic | 69.8% | Not reported | Unknown |
| 26 | Claude Sonnet 4.6Anthropic | 69.5% | Not reported | Closed |
| 27 | gpt-5.2 (xhigh)OpenAI | 67.6% | Not reported | Unknown |
| 28 | kimi-k2p5Kimi | 64.4% | Not reported | Unknown |
| 29 | Nemotron 3 Ultra (thinking)Nvidia | 63.1% | Not reported | Unknown |
| 30 | gemini-3-flash-previewGoogle | 62% | Not reported | Unknown |
| 31 | claude-sonnet-4-5 (thinking)Anthropic | 59.5% | Not reported | Unknown |
| 32 | glm-4p7Z.AI | 58.1% | Not reported | Unknown |
| 33 | gemini-3.1-flash-lite (high)Google | 57.1% | Not reported | Unknown |
| 34 | gpt-5.4-mini (xhigh)OpenAI | 56.7% | Not reported | Unknown |
| 35 | gpt-5.1 (high)OpenAI | 50.1% | Not reported | Unknown |
| 36 | o3-proOpenAI | 44.5% | Not reported | Closed |
| 37 | Claude Haiku 4.5Anthropic | 40.2% | Not reported | Closed |
How to read this leaderboard
Compare the published configurations as complete evaluation systems. The source can combine a base model, agent scaffold, tools, budget, and inference setting in each result.
Operator receipt: 37 sourced rows are currently displayable on this page; the leading published row is Gemini 4 Argon at 90.7%.
Honest limit: This Scale table is display-only context, not benchmark provenance or a weighted model-only comparison.
How BenchLM shows MCP Atlas
BenchLM mirrors 37 published rows from Scale Labs’ public MCP Atlas leaderboard, captured on November 12, 2026 snapshot.
The table is display only. It is useful context for a published agent or model configuration, but it does not enter BenchLM’s overall or category rankings.
The published MCP Atlas snapshot places Gemini 4 Argon first at 90.7%. The third row is 3.5 points behind. The broader top-10 range is 7.4 points, so many of the published results sit in a relatively narrow band.
37 models have been evaluated on MCP Atlas. The benchmark falls in the Agentic category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. MCP Atlas is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About MCP Atlas
Year
2026
Tasks
37 published rows
Format
Published Scale leaderboard score
Difficulty
External agent and model evaluation
BenchLM mirrors 37 published rows from the MCP Atlas public table captured on November 12, 2026 snapshot.
Freshness and provenance
Version
MCP Atlas 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 MCP Atlas measure?
A Scale Labs public leaderboard mirrored as display-only reference data. It does not affect BenchLM rankings.
Which model leads the published MCP Atlas snapshot?
Gemini 4 Argon currently leads the published MCP Atlas snapshot with 90.7% scale score. BenchLM shows this benchmark for display only and does not use it in overall rankings.
How many models are evaluated on MCP Atlas?
The November 12, 2026 snapshot snapshot contains 37 AI models.
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.