Vals Inference Provider - First Party - Vibe Code Bench: leaderboard
Fixed165-task challenge set;33 selected tasks for Vibe Code Bench. Same agent and tasks; provider recommended output budgets retained per row. Finance averages nine categories; overall gives three benchmarks equal weight. Separate provider cohorts; no latency or cost score. Higher accuracy is better. Endpoint provenance is retained; provider infrastructure does not create a new model identity.
Metric: Accuracy (%). Source: www.vals.ai. Saturation forecast: Around January 2027. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V4.1 Flash (High) | 85.49 | #34 (DeepSeek V4.1 Flash) |
| 2 | Kimi K3 (Max) | 80.13 | #32 (Kimi K3) |
| 3 | MiniMax-M3 | 55.23 | #147 |
No result here: #4 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vals-inference-provider-first-party-vibe-code-bench · How It Works · Data refreshed daily, snapshot 2026-10-24.