19 Sources
[1]
AI companies are now racing to the bottom -- crashing token prices and competitive models push companies to cut costs
For the rest of us, there's now significant AI intelligence to be found in more affordable models. We've entered a new phase of the AI industry's development, with all the major players heavily cutting costs and boosting the capabilities of their entry-level models in order to compete with new
[2]
China turns up the heat with open model blitz as US model makers panic
The AI arms race reached a fever pitch on Monday after Chinese e-commerce and cloud provider Alibaba called into question America's technological lead with the launch of Qwen 3.8-Max, a 2.4 trillion-parameter model that goes toe-to-toe with the best models from Anthropic and OpenAI. The new model
[3]
OpenAI drops GPT-5.6 Luna and Terra API prices by up to 80%
The lower prices are more likely to accelerate enterprise AI deployments than reduce CIO budgets, analysts say. OpenAI has cut API prices for its GPT-5.6 Terra and Luna models by 20% and 80%, respectively, while also reducing the number of usage credits the models consume in ChatGPT Work and
[4]
OpenAI cuts prices on smaller models as businesses scrutinize AI spend
July 30 (Reuters) - OpenAI slashed prices of its low- and mid-tier AI models on Thursday, a move that may intensify competition in the industry as U.S. companies battle cheaper Chinese rivals for customers increasingly wary of the technology's ballooning costs. The ChatGPT maker lowered the cost
[5]
OpenAI cuts prices for two of its GPT-5.6 AI models as companies grow sensitive to costs
OpenAI on Thursday announced it is slashing the price of two of its latest artificial intelligence models, GPT-5.6 Terra and GPT-5.6 Luna, roughly three weeks after their public release. The company is facing pressure to cater to a more cost-sensitive customer base, where enterprises have been
[6]
Qwen3.8-Max arrives with a bold claim: it outperforms GPT-5.6 Sol Max and Fable 5 on agentic computer use
Chinese e-commerce and cloud giant Alibaba's famed Qwen team of AI researchers last night unveiled Qwen3.8-Max, a new flagship 2.4-trillion-parameter mixture-of-experts (MoE) multimodal large language model (LLM) that targets one of the most competitive corners of the frontier AI market: autonomous
[7]
DeepSeek just slashed AI prices again, and China's AI race is getting even messier
China's AI race has never really been short on competition, but it's increasingly looking like a battle over who can charge the least rather than who can build the best model. DeepSeek has unveiled its latest open-weight AI model, called V4-Flash, alongside a dramatic price reduction that makes it
[8]
AI price wars: OpenAI cuts GPT-5.6 Luna prices by 80% as model competition shifts toward cost
To quote an ancient Jedi Master "Begun, the AI price wars have!" OpenAI is sharply reducing the prices of two models in its GPT-5.6 frontier series, cutting GPT-5.6 Luna, the smallest and fastest model in the series, by 80% and GPT-5.6 Terra, the mid-tier model, by 20%, while adding a premium Fast
[9]
OpenAI cuts prices on smaller models as businesses scrutinise AI spend
The ChatGPT maker lowered the cost of its smaller GPT-5.6 Luna model by 80% and its mid-tier Terra by 20%, while leaving the price of its biggest and flagship Sol model unchanged. OpenAI slashed prices of its low- and mid-tier AI models on Thursday, a move that may intensify competition in the
[10]
OpenAI Slashes GPT 5.6 Luna API Pricing by a Massive 80%
OpenAI's recent moves are reshaping the AI landscape, with significant price cuts and performance upgrades to its GPT 5.6 models. Notably, the GPT 5.6 Luna model has seen an 80% price reduction, made possible by advancements in the GPT 5.6 Sol infrastructure, which enhances efficiency without
[11]
OpenAI Goes After The Jugular Of China's Open-Weight AI Models, Cuts Token Prices By Up To 80%
Let the AI price wars begin. OpenAI has just fired a veritable volley across the bow of China's growing number of AI labs, sacrificing its sky-high margins to try to starve the budding open-weight AI economy. OpenAI is now trying to starve out China's AI labs by cutting token prices by as much as
[12]
QUICK SPARK: OpenAI Cuts AI Model Prices as Businesses Push Back on Rising Costs
OpenAI is cutting prices on some of its artificial intelligence models as businesses grow increasingly cautious about rising AI expenses and competition intensifies across the industry. The ChatGPT maker lowered the cost of its smaller GPT-5.6 Luna model by 80% and reduced pricing for its mid-tier
[13]
OpenAI Cuts Prices on Select Models to Make High-Volume Work Economical | PYMNTS.com
The company said in a Thursday blog post that it made these changes to improve the models' performance per dollar across enterprise workloads. Users can select the right model for the outcome they seek, balancing the stakes, cost of error, urgency and scale of each workflow, and the changes to
[14]
OpenAI cuts GPT-5.6 Luna pricing by 80%, introduces Fast mode for Sol API
OpenAI has announced pricing reductions for its GPT-5.6 model family, reducing the cost of GPT-5.6 Luna by 80% and GPT-5.6 Terra by 20%. The company says the changes are based on improvements across models, inference systems and agent workflows to reduce the cost and time required for AI
[15]
OpenAI sharply cuts prices for some AI models to accelerate adoption
OpenAI cut the price of its GPT-5.6 Luna model by 80% and its midtier Terra model by 20%, while keeping pricing for its top-end Sol model unchanged. Companies will now pay 20 cents per million input tokens for Luna, down from $1 previously, while the cost to generate responses falls from $6 to
[16]
OpenAI sharply cuts prices for some AI models to speed adoption
OpenAI has cut the price of its GPT-5.6 Luna model by 80% and its mid-tier Terra model by 20%, while keeping the price of its top-end Sol model unchanged. Companies will now pay 20 cents per million input tokens for Luna, down from $1 previously, while the cost to generate responses falls from $6
[17]
OpenAI cuts prices on smaller models as businesses scrutinize AI spend
July 30 (Reuters) - OpenAI slashed prices of its low- and mid-tier AI models on Thursday, a move that may intensify competition in the industry as U.S. companies battle cheaper Chinese rivals for customers increasingly wary of the technology's ballooning costs. The ChatGPT maker lowered the cost
[18]
The great AI price war: How Chinese open models are squeezing GPT-5.6 Sol and Claude Fable 5
The age when the frontiers of artificial intelligence research belonged to select American laboratories was one of great monopoly, where OpenAI and Anthropic set the trend and price standards, and others in the industry followed suit. All that has changed in July 2026, with two Chinese laboratories
[19]
OpenAI makes GPT-5.6 dramatically cheaper, putting pressure on Anthropic and rivals
The revised pricing undercuts Anthropic's Claude Sonnet 4.6 and comes amid growing pressure to lower AI deployment costs for businesses. OpenAI has announced a massive price cut for its GPT 5.6 family of AI models lowering API costs by as much as 80 pct for some offerings. This move is said to
Share
Copy Link
OpenAI has dramatically reduced pricing for its GPT-5.6 Luna and Terra models by 80% and 20% respectively, responding to mounting pressure from Chinese competitors like DeepSeek and Alibaba. The move signals a fundamental shift in AI economics as companies prioritize affordability alongside capability.
OpenAI announced sweeping AI price cuts on July 30, slashing costs for its GPT-5.6 Luna model by 80% and GPT-5.6 Terra by 20%
3
4
. The aggressive pricing strategy represents a dramatic shift from March 2026, when GPT-5.4 launched at $2.50 per million input tokens and $15 per million output tokens1
. GPT-5.6 Luna now costs just $0.20 per million input tokens and $1.20 per million output tokens, while Terra drops to $2 and $12 respectively3
. This means frontier AI models remain cutting-edge for less than four months before pricing collapses1
. The flagship GPT-5.6 Sol maintains its $5 and $30 pricing, though a new Fast mode charges $10 and $60 per million input/output tokens for lower latency1
.
Source: VentureBeat
Chinese competitors have fundamentally altered the AI competition landscape. DeepSeek V4 Flash costs $0.435 per million input tokens and $0.87 per million output tokens, while Moonshot's Kimi K3 charges $3.00 and $15.00 respectively
1
. According to independent benchmarks by Artificial Analysis, DeepSeek V4 Flash 0731 performs within a single point of GPT-5.6 Luna while costing 40% less per task2
. Alibaba escalated competition further by releasing Qwen 3.8-Max, a 2.4 trillion-parameter model that matches capabilities from Anthropic and OpenAI2
. Hugging Face CEO Clément Delangue stated on CNBC that China is "clearly dominating on open models right now," predicting they could dominate at the frontier by year-end2
. These open-weight models from DeepSeek, Alibaba, Moonshot, MiniMax, and Z.ai provide enterprises their only credible alternatives to proprietary systems2
.
Source: The Register
Businesses have grown increasingly wary of AI spending as costs balloon without clear returns on investment
5
. Companies using AI heavily have complained about skyrocketing token costs for months, making cheaper alternatives particularly attractive1
. The shift from flat subscriptions to usage-based pricing leaves enterprises with unpredictable and often higher bills as usage per task becomes harder to estimate4
. OpenAI attributed the lower prices partly to efficiency gains from GPT-5.6, including improved code optimization during internal development4
. Analysts suggest these reductions will more likely accelerate enterprise AI deployments rather than reduce CIO budgets3
. Google introduced more-affordable Gemini 3.6 Flash and 3.5 Flash-Lite models, while Anthropic replaced its Opus 4.8 with a more capable Claude 5.0 at the same price point1
.Related Stories
OpenAI faces mounting financial pressures despite these competitive moves. The company is already losing money on subscription-based accounts and missed key revenue targets earlier this year after losing tens of billions in 2025
1
. OpenAI has committed to $600 billion in compute spend by 2030, including a $300 billion compute commitment with Oracle1
. Cutting prices on the most popular, affordable models suggests margins will either shrink dramatically or disappear altogether1
. Reports indicate Nvidia may backstop OpenAI with a $250 billion investment1
. Google spent approximately nine times its cloud revenue on AI infrastructure over the past year, while Anthropic only recently posted profits on annualized revenue through a limited cut-price deal with xAI to rent its Colossus data center1
. Analysts warn that cutting prices could boost usage but strain finances ahead of highly anticipated initial public offerings4
.The new pricing intensifies pressure on Anthropic, whose Claude Sonnet 4.6 model costs $3 per million input tokens and $15 per million output tokens, above Terra's rates
4
. Anthropic CEO Dario Amodei recently stated opposition to open models from China, ones distilled from proprietary models, and those not meeting rigorous safety metrics2
. However, neither safety concerns nor commitments from American and European tech giants change the reality that China provides the only meaningful competition in the open-weight models arena2
. Alibaba's Qwen 3.8-Max is now available via QwenCloud for $2 per million input tokens and $6 per million output tokens, with open weights released for the first time2
. Despite more workers using AI than ever before, productivity gains have been less than ideal, prompting major AI businesses to announce cuts and limits on technology use1
. Watch how OpenAI balances its massive compute costs commitments against shrinking margins, whether Anthropic adjusts pricing to remain competitive, and if Chinese developers continue expanding their open-weight model advantage.
Source: Tom's Hardware
Summarized by
Navi
[1]
26 Jul 2026•Policy and Regulation

24 May 2026•Technology

17 Jun 2026•Technology

1
Science and Research

2
Policy and Regulation

3
Technology