# Vals Router Index (Vals AI) (Vals Router Index)

> Vals AI comparison of model routers with single models on long agentic work in finance, coding, law and tax, using the same agents, tasks and graders as the Vals Index.

Canonical page: https://benchlm.ai/benchmarks/vals-router-index

- Category: [Agentic](/agentic)
- Last updated: October 20, 2026

Evaluation results published by [Vals AI](https://www.vals.ai/benchmarks/vals_router_index). Published leaderboard aggregates and evaluation settings only. Dataset contents are not included.

## About Vals Router Index

- Year: 2026
- Tasks: About 1,250 agentic tasks
- Format: Accuracy, cost per test and duration
- Difficulty: Long tool-using agent work in finance, coding, law and tax
- Paper: [Vals Router Index leaderboard and methodology](https://www.vals.ai/benchmarks/vals_router_index)

Vals published the index on October 20, 2026. Five routers (OpenRouter Auto, Fireworks FireRouter, Ramp Router, TypeSafe Jev Router and Sakana Fugu Max) run about 1,250 tasks from the eight agentic benchmarks of the Vals Index, six of them private, each in its default configuration. Vals reports accuracy, cost per test and duration and compares each router with single models. We mirror the published aggregates as a display-only board, outside overall and category rankings.

Vals Router Index is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (27 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [Gemini 4 Argon](/models/gemini-4-argon) | Google | 69.24% |
| 2 | [Claude Sonnet 5.5](/models/claude-sonnet-5-5) | Anthropic | 69.03% |
| 3 | [Muse Wonder Full](/models/muse-wonder-full) | Meta | 66.79% |
| 4 | [GPT-6 Astra](/models/gpt-6-astra) | OpenAI | 65.02% |
| 5 | [Muse Wonder](/models/muse-wonder) | Meta | 64.88% |
| 6 | [Claude Opus 5.5](/models/claude-opus-5-5) | Anthropic | 64.16% |
| 7 | [GPT-6.1 Sol](/models/gpt-6-1-sol) | OpenAI | 63.94% |
| 8 | [DeepSeek V4.1 Flash 1017](/models/deepseek-v4-1-flash) | DeepSeek | 61.26% |
| 9 | [GLM-5.3](/models/glm-5-3) | Z.AI | 59.73% |
| 10 | [Claude Haiku 5.5](/models/claude-haiku-5-5) | Anthropic | 58.71% |
| 11 | [Grok 4.7](/models/grok-4-7) | xAI | 58.56% |
| 12 | [MiMo-V2.6-Pro](/models/mimo-v2-6-pro) | Xiaomi | 57.81% |
| 13 | [Ramp Router](https://www.vals.ai/models/ramp_switchyard) | ramp | 57.80% |
| 14 | [Fireworks FireRouter](https://www.vals.ai/models/fireworks_firerouter) | Fireworks | 56.46% |
| 15 | [Gemini 3.8 Flash](/models/gemini-3-8-flash) | Google | 55.95% |
| 16 | [TypeSafe Jev Router](https://www.vals.ai/models/openrouter_typesafe/jev-router) | openrouter | 55.65% |
| 17 | [MiMo-V2.6-Flash](/models/mimo-v2-6-flash) | Xiaomi | 54.83% |
| 18 | [MiniMax M3.1 Flash](/models/minimax-m3-1-flash) | MiniMax | 53.78% |
| 19 | [Hy4 preview](/models/hy4-preview) | Tencent | 53.21% |
| 20 | [DeepSeek V4 Pro 0813](/models/deepseek-v4-pro-0813) | DeepSeek | 52.97% |
| 21 | [Kimi K3](/models/kimi-k3) | Moonshot AI | 52.95% |
| 22 | [GPT-6 Luna](/models/gpt-6-luna) | OpenAI | 52.32% |
| 23 | [OpenRouter Auto](https://www.vals.ai/models/openrouter_openrouter/auto) | openrouter | 52.20% |
| 24 | [GLM-5.3-Flash](/models/glm-5-3-flash) | Z.AI | 50.86% |
| 25 | [Sakana Fugu Max](https://www.vals.ai/models/sakana_fugu-max-v1.0) | sakana | 48.11% |
| 26 | [Inkling](/models/inkling) | Thinking Machines Lab | 28.96% |
| 27 | [Inkling-Small](/models/inkling-small) | Thinking Machines Lab | 27.59% |

## FAQ

### What does Vals Router Index measure?

Vals AI comparison of model routers with single models on long agentic work in finance, coding, law and tax, using the same agents, tasks and graders as the Vals Index.

### Which model leads the published Vals Router Index snapshot?

Gemini 4 Argon currently leads the published Vals Router Index snapshot with a score of 69.24%.

### How many models are evaluated on Vals Router Index?

The October 20, 2026 contains 27 AI models.

### Does Vals Router Index affect BenchLM's overall score?

Not directly. Vals Router Index is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.