Breaking

OpenAI and Anthropic Price Cuts Put New Pressure on Open-Weight AI's Cost Edge

September 26, 2026 at 19:01 EDT

Foundation Models
Open Source
AI Agents

OpenAI's newest low-cost GPT-6 models are tightening the price gap that made open-weight AI the default choice for many high-volume developers. The company released GPT-6 Sol and GPT-6 Luna on September 22 with API prices 50% below GPT-5.6 promotional rates, putting Luna at $0.10 per 1 million input tokens and $0.50 per 1 million output tokens, according to OpenAI's launch details.

Sep 2026 · AI pricing

Closed APIs close the cost gap

OpenAI's GPT-6 Luna puts a managed proprietary model into the same low-cost band as leading hosted open-weight options, shifting the decision from sticker price to workload economics.

50%

Lower API prices for GPT-6 Sol and Luna versus GPT-5.6 promotional pricing

$0.50

GPT-6 Luna output price per 1M tokens

79%

Developers adding AI features who use open models

51% vs 63%

Open-model production rate versus closed-model production rate

Output price comparison

Per 1M output tokens. Each block represents $0.10.

$0.50

GPT-6 Luna

$0.60

DeepSeek Flash off-peak

Takeaway: the headline hosted-output price gap is now only $0.10 per 1M tokens for these two low-cost options.

The new routing logic

Lower list price

→

More agent loops

→

Benchmark cost per task

→

Route by workload

Why managed APIs regain ground

Hosted infrastructure, tool support, caching controls and enterprise reliability come bundled, making high-volume agents easier to run.

Why open weights stay sticky

Local deployment, data control, custom tuning and very cheap cache-heavy inference still matter when teams need ownership and portability.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.