# Best Value LLM for Math in 2026 — Cost-Adjusted Rankings

> Top AI models ranked by math benchmark performance per dollar. Cost-adjusted rankings using AIME 2025, BRUMO, and MATH-500.

This ranking divides each model's weighted math score (AIME, BRUMO, MATH-500) by its output token price, so it shows which models give the most math capability per dollar. A cheap model with a low math score can rank high here, so read the score column next to the value column before you shortlist one.

Canonical page: https://benchlm.ai/best/best-value-math

Last updated: October 8, 2026

## Rankings

| Rank | Model | Creator | Type | Context | Score/$ | Score | Output $/1M |
|------|-------|---------|------|---------|---------|-------|-------------|
| 1 | [Kimi K2.5](/models/kimi-k2-5) | Moonshot AI | Open Weight | 256K | 20.7 | 62.1 | $3 |
| 2 | [Kimi K2.6](/models/kimi-2-6) | Moonshot AI | Open Weight | 256K | 17.75 | 71 | $4 |
| 3 | [GLM-5](/models/glm-5) | Z.AI | Open Weight | 200K | 17.66 | 56.5 | $3.2 |
| 4 | [GLM-5.1](/models/glm-5-1) | Z.AI | Open Weight | 203K | 14.5 | 63.8 | $4.4 |
| 5 | [Claude Opus 4.8](/models/claude-opus-4-8) | Anthropic | Proprietary | 1M | 2.61 | 65.2 | $25 |
| 6 | [Claude Opus 4.5](/models/claude-opus-4-5) | Anthropic | Proprietary | 200K | 2.32 | 57.9 | $25 |

## Key Takeaways

- Top model: [Kimi K2.5](/models/kimi-k2-5) with a Score/$ of 20.7 (score 62.1, $3/1M output tokens)
- Best open-weight option: [Kimi K2.5](/models/kimi-k2-5) at #1
- Models included: 6

## Compare the Leaders

- [Kimi K2.5 vs Kimi K2.6](/compare/kimi-2-6-vs-kimi-k2-5)