A lower price per litre doesn’t help much if your new car drinks three times as much. That’s the AI token price cut OpenAI is reportedly weighing, in one line. Your invoice is price per token times the tokens your team burns, and agent tools grow the second number faster than any vendor can shrink the first. Budget by what a finished task costs.
Key points
- OpenAI is reportedly considering major token price cuts. No numbers, dates or models yet.
- In our hypothetical, a 30% cut plus agents that double tokens per task turns a $10,000 bill into $14,000.
- Rank teams by finished work. A usage leaderboard buys usage.
- Three numbers decide whether a cut matters to you: last month’s spend, finished tasks, and who can say stop.
What’s actually been reported about the AI token price cut
Reports on June 11 said OpenAI is considering major cuts to token prices and expects similar moves from Anthropic. OpenAI has not published figures. Treat it as a reported plan you cannot budget against yet. Sam Altman has called token costs a huge issue for enterprise customers, and blowing through a quarterly token budget has become nearly a meme. The vendor sees churn risk. You should see a budgeting problem its pricing page will not fix.
The math that eats a price cut
A chat prompt uses a small, predictable number of tokens. A coding agent that reads a codebase, tries a fix, fails and tries again can use many times more per task. Cheaper units invite more units. That’s our analysis, and nobody has measured it for your firm.
Our hypothetical, step by step:
- A team spends $10,000 a month and merges 200 pull requests. $50 per finished task.
- A 30% cut would bring the same work to $7,000.
- The team moves to agents that loop, and tokens per task double. The bill lands at $14,000.
- It merges 260 pull requests. Output up 30%, spend up 40%, about $54 per finished task. The AI token price cut got the credit.
Nobody sees that without counting the pull requests. For steady usage, a cut is plain good news: same volume, 30% less, and a price war between two big vendors tends to help buyers over a year or two. Take the saving. Keep it out of the forecast until tokens per task have held still for a quarter. Hardware works the same way, where robots vs human labour cost turns on usage rather than the sticker.
Uber’s leaderboard, and a better one
Uber’s COO, Andrew Macdonald, has said it is hard to connect Claude Code usage to useful consumer features. The link between usage and useful features is still missing. Uber used up its 2026 AI coding budget in 4 months, with an internal leaderboard ranking teams by use. A similar claim has been attributed to Uber’s CTO in April, so the source of the 4-month figure is unsettled. The arithmetic above holds either way.
Rank teams by finished work. Merged pull requests, closed tickets and approved drafts per team per month cost nothing to count, and they reward the team that finishes on fewer tokens. Celebrate tokens burned and you’re running a spending contest the vendor wins. Uber can afford to learn that in public. A 50-person firm with one finance lead learns it in a quarterly review.
Three numbers, one line a month
Last month’s spend sits on the invoice. Finished tasks and the name of whoever can say stop usually aren’t written anywhere. Post all three as one line in the finance channel each month; three months make a trend. Then name the workflow that burns the most tokens, decide what you’d drop if its budget halved, and confirm a spending cap exists with an owner.
What AI actually costs a Canadian business covers forgotten line items, and the hidden time tax of AI work covers cleanup hours no invoice shows. Anthropic’s Opus 4.8 price hold is the same lesson from the other vendor, Ramp’s AI Index shows who’s being billed, and our look at GPT-5.5 token efficiency shows why price is half the story. A cheap token buys nothing if the model makes things up, which is why Jev’s no-writing design is worth a look.
If a cut arrives, ask your vendor in writing whether your contract passes a lower list price through. Committed-spend deals can lag the public page. Real figures from OpenAI, per-task reporting or hard spend caps would make price matter more than we’ve allowed. Until then, pull last month’s invoice and count the finished tasks.
Frequently asked questions
Is OpenAI cutting its token prices?
Reportedly, yes. OpenAI is reported to be considering major cuts, but the company has not announced figures, dates or models, so it is not a confirmed price change.
Will cheaper tokens lower my company’s AI bill?
Only if your usage stays the same. Agent and coding tools can use many times more tokens per task than chat, so a unit price cut can be outrun by volume.
How should a small business budget for AI usage?
Track cost per finished task, set a spending cap with a named owner, and review the biggest workflow each month. Token price alone tells you very little.