nAIvigate · Barometer
The AI model market in September 2026
157 models tracked. Figures computed on September 11 from verified model sheets and the daily history. Free to reuse with attribution (CC BY 4.0).
Daily history available since September 7: this month's variations cover a partial window.
Trust Index: top 5
- 1. Mistral Large 398/100
- 2. StarCoder 296/100
- 3. FLUX.2 [klein] 4B96/100
- 4. Magistral Small96/100
- 5. Mistral Nemo 12B96/100
Cheapest (input + output)
- 1. Yi-Lightning0.14 / 0.14 $
- 2. DeepSeek Coder V30.14 / 0.28 $
- 3. DeepSeek V3.20.28 / 0.42 $
- 4. Mistral Small 30.2 / 0.6 $
- 5. Mixtral 8x22B0.6 / 0.6 $
Method
Prices in US dollars per million tokens, collected from vendor pricing pages and OpenRouter, median over models with a verified public price. Trust Index: 16 documented criteria in 5 families, an undocumented criterion scores 0, only models with 50%+ coverage enter the average. Variations: first and last daily snapshot of the month (384 snapshots, September 7 to September 10). No placement is sold, no figure is produced by an LLM.
Cite: "nAIvigate Barometer, September 2026", https://naivigate.com/en/barometre/2026-09