Issue 2
Welcome to AI HQ Daily
A short, factual briefing on what changed in the model market, built from live listings rather than press releases. Here is how it works and what to expect.
Most AI newsletters are written the night before and are out of date by the time they land. AI HQ Daily takes a different approach: the words are ours, but every number on this page is pulled from provider listings when you open it. If a price changes or a model appears after we publish, the figures below move with it.
Each issue is deliberately short. One or two paragraphs of context, a live view of what is new, and the numbers that matter for anyone paying per token. No hype, no leaderboards we cannot back up, and no scores we invented ourselves.
New on the listings this week
25 models first listed in the last 7 days; showing the latest 12. Full Release Radar
- Fireworks: Ember-1 Fireworks · 24 Sept 2026$3.00 in · $15.00 out
- Z.ai: GLM 5.3 Prime Z.ai (Zhipu) · 23 Sept 2026$2.80 in · $8.80 out
- Qwen: Qwen3.8 Max Prime Qwen (Alibaba) · 23 Sept 2026$4.00 in · $12.00 out
- AionLabs: Aion 3.5 Mini Aion Labs · 23 Sept 2026$0.700 in · $1.40 out
- AionLabs: Aion 3.5 Aion Labs · 23 Sept 2026$3.00 in · $6.00 out
- Upstage: Solar Mini 4 Upstage · 23 Sept 2026$0.050 in · $0.200 out
- Cohere: Command A+ Cohere · 22 Sept 2026$0.300 in · $1.50 out
- OpenAI: GPT-6 Luna Pro (batch) OpenAI · 22 Sept 2026$0.050 in · $0.250 out
- OpenAI: GPT-6 Luna Pro OpenAI · 22 Sept 2026$0.100 in · $0.500 out
- OpenAI: GPT-6 Luna (batch) OpenAI · 22 Sept 2026$0.050 in · $0.250 out
- OpenAI: GPT-6 Luna OpenAI · 22 Sept 2026$0.100 in · $0.500 out
- OpenAI: GPT-6 Sol Pro (batch) OpenAI · 22 Sept 2026$1.00 in · $5.00 out
How to read the release list
The dates shown are when a model first appeared on the aggregated listing we track, which is usually within a day of the vendor announcement but is not the vendor's own release date. Pricing is the list rate for the primary route; cached-input and batch discounts are not shown here.
A flurry of small releases from one provider often means variants of the same base model (different context sizes, speed tiers or regions) rather than genuinely new capability. Check the model page before drawing conclusions.
The market in numbers
- Models listed
- 431
- Providers
- 58
- Listed in last 30 days
- 57
- Cheapest input
- $0.017/M
- Cheapest output
- $0.030/M
- Median blended
- $0.850/M
- 431 paid models
Source: OpenRouter (live)
Where the flagships sit today
Resolved automatically to the newest paid model in each family.
OpenAI
$0.450
/M blended
OpenAIAnthropic
$8.00
/M blended
Anthropic$4.50
/M blended
Google$2.40
/M blended
xAIDeepSeek
$0.262
/M blended
DeepSeek
What comes next
Issues will appear on weekdays when there is something worth saying, and the archive stays here so you can see how the picture changed. Subscribe below if you would rather have it arrive by email; the newsletter version uses the same blocks, frozen at send time.