Themed ranking
Best value
Ranked by quality per dollar - Artificial Analysis Intelligence Index divided by blended price per 1M tokens. A plain ratio, no invented weighting.
As of September 23, 2026, inclusionAI: Ling 3.0 Flash ranks #1 for best value at quality per $ 654.0 pts/$.
- #1 inclusionAI: Ling 3.0 Flash - Quality per $ 654.0 pts/$
- #2 inclusionAI: Ling 3.0 Flash VL (free) - Quality per $ 273.3 pts/$
- #3 inclusionAI: Ling 3.0 Flash Fin (free) - Quality per $ 251.1 pts/$
- #4 OpenAI: gpt-oss-20b (batch) - Quality per $ 250.0 pts/$
- #5 DeepSeek: DeepSeek V4.1 Flash (batch) - Quality per $ 197.5 pts/$
- #6 OpenAI: GPT-6 Luna (batch) - Quality per $ 186.5 pts/$
- #7 DeepSeek: DeepSeek V4 Flash 0423 - Quality per $ 180.5 pts/$
- #8 Z.ai: GLM 5.3 Flash (batch) - Quality per $ 176.0 pts/$
- #9 NVIDIA: Nemotron 3.5 Lightning (free) - Quality per $ 125.9 pts/$
- #10 IBM: Granite 4.2 8B - Quality per $ 103.3 pts/$
Top 15 by quality per dollar
This is a ratio, not a capability ranking: an ultra-cheap model with modest absolute quality can outscore a much stronger, pricier one, purely because price sits near zero. Each bar shows its own Quality and Price/1M alongside the ratio. The highlighted bar is today's single best value.
How this is calculated
Quality per $ = Intelligence Index ÷ blended price per 1M tokens - a plain ratio, not a capability ranking. An ultra-cheap model with modest absolute Quality can outrank a much stronger, pricier one purely because price sits near zero; the Quality and Price/1M columns above show the two real numbers behind every ratio.
What it can't tell you: consistency run-to-run, how costly a wrong answer is for your use case, or whether a task needs multiple attempts. For high-stakes work, weigh the absolute Quality score more than the ratio, and check measured endpoint uptime for consistency.
Source: OpenRouter (openrouter.ai/rankings), as of September 23, 2026.
Benchmark scores: Artificial Analysis (artificialanalysis.ai) via OpenRouter (openrouter.ai/rankings).
Methodology: www.smophy.ai/benchmark/methodology
Token counts originate from each provider's own tokenizer and are not directly comparable across providers.
