Skip to content
Model Season every model has a seasonRadar
Model Season

Radar

What changed in language models, one day at a time: what is in the news, what entered and left the catalog, what changed price, what is being deprecated and what moved in traffic. Each fact comes with what the usage history says about it.

Edition of Sep 16, 20263 editionsRSS
Traffic · Xiaomi

MiMo-V2.5 gained 3.6 points of share

From 2.8% to 6.4% between the last two 7-day windows.

What the data says. Xiaomi has 6.6% of tokens and 1.1% of estimated spend this week.

Price · Z.ai (GLM)

GLM 5.2 got 126% more expensive

Blended price went from $0.95 to $2.15 per 1M tokens (input $0.6 to $1.40, output $2.00 to $4.40), unchanged since Sep 15, 2026.

What the data says. The model has 1.19% of traffic over the last 7 days, ranked 15th. Z.ai (GLM) has 13.6% of tokens and 8.6% of estimated spend this week. The lab's most used model is GLM 5.3 Flash, with 9.17% at $0.142 per 1M, and this one costs 15 times more.

Price · DeepSeek

DeepSeek V4 Flash 0423 got 33% more expensive

Blended price went from $0.084 to $0.111 per 1M tokens (input $0.067 to $0.089, output $0.134 to $0.177), unchanged since Sep 15, 2026.

What the data says. The model has 3.47% of traffic over the last 7 days, ranked 7th. DeepSeek has 20.2% of tokens and 5.6% of estimated spend this week. The lab's most used model is DeepSeek V4 Flash 0731, with 9.24% at $0.075 per 1M, and this one costs 1.5 times more.

Price · Qwen

Qwen3.5-35B-A3B got 18% cheaper

Blended price went from $0.547 to $0.447 per 1M tokens (input $0.313 to $0.163, output $1.25 to $1.30), unchanged since Sep 15, 2026.

What the data says. No volume over the last 7 days. Its weekly peak was 0.39%. The lab's most used model is Qwen3.8 Max (0902), with 0.27% at $3.00 per 1M, and this one costs 6.7 times less.

Catalog snapshot of Sep 16, 2026, compared with the previous ones since Sep 12, 2026. Traffic through Sep 15, 2026. Price is a 75% input and 25% output blend, and only becomes news when the new value holds for two snapshots. Order follows a fixed relevance formula that weighs the type of change and the model’s share.

Permanent link to this edition

Previous editions