Vals EMB (EMB)
We show this table for reference; we do not rank on it.
Evaluating agents on Excel-based financial modeling tasks
Evaluation results published by Vals AI. Published leaderboard aggregates and evaluation settings only. Dataset contents are not included.
EMB score on EMB — October 12, 2026
We mirror the published emb score view for EMB. Claude Fable 5.1 leads the public snapshot at 76.67%, followed by Claude Opus 5.5 (75.94%) and Claude Sonnet 5.5 (75.71%). We do not use these results to rank models overall.
Claude Fable 5.1
Anthropic
Claude Opus 5.5
Anthropic
Claude Sonnet 5.5
Anthropic
74 modelsAgenticCurrentDisplay onlyUpdated October 12, 2026
EMB score table (74 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Claude Fable 5.1Anthropic | 76.67% | Not reported | Closed |
| 2 | Claude Opus 5.5Anthropic | 75.94% | Not reported | Closed |
| 3 | Claude Sonnet 5.5Anthropic | 75.71% | Not reported | Closed |
| 4 | Gemini 4 ArgonGooglehigh reasoning | 75.24% | Not reported | Closed |
| 5 | Claude Fable 5Anthropic | 73.67% | Not reported | Closed |
| 6 | Claude Opus 5Anthropic | 73.56% | Not reported | Closed |
| 7 | GPT-5.6 SolOpenAImax reasoning | 72.34% | Not reported | Closed |
| 8 | Gemini 3.8 FlashGooglehigh reasoning | 72.20% | Not reported | Closed |
| 9 | Muse Wonder FullMetamax reasoning | 71.92% | Not reported | Closed |
| 10 | GPT-6 AstraOpenAImax reasoning | 71.70% | Not reported | Closed |
| 11 | GPT-6 SolOpenAImax reasoning | 71.53% | Not reported | Closed |
| 12 | Gemini 3.7 FlashGooglehigh reasoning | 71.33% | Not reported | Closed |
| 13 | GPT-6.1 SolOpenAImax reasoning | 70.77% | Not reported | Closed |
| 14 | Claude Opus 4.8Anthropic | 69.37% | Not reported | Closed |
| 15 | GPT-6 LunaOpenAImax reasoning | 68.52% | Not reported | Closed |
| 16 | Muse Spark 1.3 MaxMetamax reasoning | 67.43% | Not reported | Unknown |
| 17 | GPT-5.6 LunaOpenAImax reasoning | 67.12% | Not reported | Closed |
| 18 | Muse WonderMetamax reasoning | 67.08% | Not reported | Closed |
| 19 | Grok 4.7xAIxhigh reasoning | 66.99% | Not reported | Closed |
| 20 | Kimi K3Moonshot AImax reasoning | 66.68% | Not reported | Pending |
| 21 | Claude Sonnet 5Anthropic | 66.32% | Not reported | Closed |
| 22 | GPT-5.6 TerraOpenAImax reasoning | 66.20% | Not reported | Closed |
| 23 | MiMo-V2.6-FlashXiaomi | 65.46% | Not reported | Open |
| 24 | Gemini 3.6 FlashGooglehigh reasoning | 65.41% | Not reported | Closed |
| 25 | GPT-5.5OpenAIxhigh reasoning | 64.54% | Not reported | Closed |
| 26 | Claude Opus 4.7Anthropic | 63.73% | Not reported | Closed |
| 27 | Gemini 3.5 FlashGooglehigh reasoning | 63.55% | Not reported | Closed |
| 28 | Ember-1Fireworks | 63.37% | Not reported | Closed |
| 29 | MiMo-V2.6-ProXiaomi | 62.86% | Not reported | Open |
| 30 | Grok 4.6xAIhigh reasoning | 62.73% | Not reported | Closed |
| 31 | Muse Spark 1.3Metaxhigh reasoning | 62.71% | Not reported | Closed |
| 32 | GLM-5.2Z.AI | 61.53% | Not reported | Open |
| 33 | Claude Sonnet 4.6Anthropic | 60.15% | Not reported | Closed |
| 34 | Qwen3.8 MaxAlibaba | 60.07% | Not reported | Open |
| 35 | Qwen3.8-27BAlibabaxhigh reasoning | 59.66% | Not reported | Open |
| 36 | Claude Haiku 5.5Anthropic | 59.02% | Not reported | Closed |
| 37 | Hy4 previewTencent | 58.08% | Not reported | Open |
| 38 | Kimi K2.6Moonshot AI | 57.85% | Not reported | Open |
| 39 | DeepSeek V4.1 Flash 1017DeepSeekhigh reasoning | 57.21% | Not reported | Open |
| 40 | Step 5 PreviewStepFun | 57.12% | Not reported | Open |
| 41 | Qwen3.7 MaxAlibaba | 57.04% | Not reported | Closed |
| 42 | Muse Spark 1.2Metaxhigh reasoning | 56.98% | Not reported | Closed |
| 43 | DeepSeek V4 Flash 0731DeepSeekhigh reasoning | 56.98% | Not reported | Open |
| 44 | Muse Spark 1.1Metaxhigh reasoning | 56.36% | Not reported | Closed |
| 45 | GLM-5.3Z.AImax reasoning | 56.34% | Not reported | Open |
| 46 | Mistral Large 4Mistralhigh reasoning | 56.21% | Not reported | Pending |
| 47 | GLM-5.3-FlashZ.AImax reasoning | 55.93% | Not reported | Open |
| 48 | MiMo-V2.5-ProXiaomi | 55.23% | Not reported | Closed |
| 49 | MiMo-V2.5Xiaomi | 55.09% | Not reported | Closed |
| 50 | Grok 4.5xAIhigh reasoning | 52.91% | Not reported | Closed |
| 51 | DeepSeek V4 Pro 0813DeepSeekmax reasoning | 52.80% | Not reported | Open |
| 52 | Gemini 3.1 Pro PreviewGooglehigh reasoning | 52.62% | Not reported | Unknown |
| 53 | DeepSeek V4 Pro 0813DeepSeekmax reasoning | 51.62% | Not reported | Open |
| 54 | Qwen3.7 PlusAlibaba | 49.34% | Not reported | Closed |
| 55 | MiniMax M3MiniMax | 47.76% | Not reported | Open |
| 56 | GPT-5.4 miniOpenAIxhigh reasoning | 45.43% | Not reported | Closed |
| 57 | GPT-5.4 nanoOpenAIhigh reasoning | 44.75% | Not reported | Closed |
| 58 | Gemini 3.5 Flash-LiteGooglehigh reasoning | 43.23% | Not reported | Closed |
| 59 | InklingThinking Machines Lab0.99 reasoning | 38.36% | Not reported | Open |
| 60 | Ling 3.0 Flash Af Rc3Ant | 36.83% | Not reported | Unknown |
| 61 | Qwen3.6 PlusAlibaba | 32.87% | Not reported | Closed |
| 62 | Inkling-SmallThinking Machines Lab0.99 reasoning | 31.76% | Not reported | Open |
| 63 | Nemotron 3 Ultra 550b A55bNvidia | 31.73% | Not reported | Unknown |
| 64 | Kimi K2.5 ThinkingMoonshot AI | 28.47% | Not reported | Unknown |
| 65 | MiniMax M2.7MiniMax | 28.45% | Not reported | Open |
| 66 | Gemini 3 Flash PreviewGooglehigh reasoning | 26.02% | Not reported | Unknown |
| 67 | Ling 3.0 Flash 2607Ant | 25.01% | Not reported | Unknown |
| 68 | Claude Haiku 4.5 ThinkingAnthropic | 22.66% | Not reported | Closed |
| 69 | Grok 4.3xAIhigh reasoning | 18.39% | Not reported | Closed |
| 70 | Nemotron Lightning 3p5 30b A3bFireworks AI | 16.99% | Not reported | Unknown |
| 71 | Grok 4.20 0309 ReasoningxAI | 11.74% | Not reported | Unknown |
| 72 | Mistral Medium 3.5Mistral AIhigh reasoning | 11.19% | Not reported | Unknown |
| 73 | Mercury 2.5Inceptionhigh reasoning | 8.79% | Not reported | Closed |
| 74 | Gemini 3.1 Flash Lite PreviewGooglehigh reasoning | 8.63% | Not reported | Unknown |
How EMB is shown here
BenchLM mirrors the public Vals AI EMB leaderboard captured from https://www.vals.ai/benchmarks/emb and updated by Vals on October 12, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.
EMB is display only on BenchLM. Vals proprietary or Vals-hosted aggregate views are useful context, but BenchLM does not use them as weighted ranking inputs or as a replacement for benchmark-native source records.
Snapshot
The published EMB snapshot places Claude Fable 5.1 first at 76.67%. The third row is 0.96 points behind. The broader top-10 range is 4.97 points, so many of the published results sit in a relatively narrow band.
74 models have been evaluated on EMB. 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. EMB is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About EMB
Year
2026
Tasks
Excel-based financial modeling tasks
Format
Accuracy score
Difficulty
Professional finance modeling
BenchLM mirrors the public Vals AI EMB leaderboard as display-only external evidence. The captured snapshot preserves overall scores, task-level scores where Vals publishes them, uncertainty, latency, and cost-per-test metadata. It is excluded from BenchLM weighted rankings.
Freshness and provenance
Version
EMB 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 EMB measure?
Evaluating agents on Excel-based financial modeling tasks
Which model leads the published EMB snapshot?
Claude Fable 5.1 currently leads the published EMB snapshot with 76.67% emb score. BenchLM shows this benchmark for display only and does not use it in overall rankings.
How many models are evaluated on EMB?
The October 12, 2026 snapshot contains 74 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 5,600+ readers.
One email each week. Unsubscribe anytime.