On June 17, 2026, Google cut prices on its Gemini 2.5 Pro API. The input price dropped from $1.25 to $0.75 per million tokens, a 40 percent cut. Output price fell from $3.75 to $2.50 per million tokens, a 33 percent cut. Google published the new rates on its official AI pricing page. The change hit developers using pay-as-you-go API access immediately. It did not alter free tier limits for consumer Gemini apps. But the move sent a clear signal. Google wanted volume. The company had already cut subscription prices earlier in June. Now it targeted API customers. The cuts put pressure on OpenAI and Anthropic to respond. Free AI News first reported the Google subscription cuts in Google AI price cuts should make OpenAI Anthropic nervous.
The price cut hit paying API users first. Developers on the Gemini free tier saw pro model access restricted for some endpoints in early June. Those restrictions are detailed in Google Gemini API free tier tightened pro models now paid. Under the new API rates, a workload of 100 million input tokens and 10 million output tokens cost $100. Before June 17, that same workload cost $187.50. The 46 percent total cost reduction was the largest single API price cut from a major provider in 2026. Google did not reduce rate limits. But the lower unit cost made Gemini 2.5 Pro competitive with smaller open models on price. Mistral and Meta had already priced open-weight models lower, but they lacked Google’s distribution and latency.
OpenAI felt the pressure immediately. On June 17, two enterprise account managers told customers the company was considering matching cuts to GPT-5.1 API rates. Current list prices were $3.00 per million input tokens and $12.00 per million output tokens. A proposed cut would bring input to $1.80 and output to $7.20. OpenAI’s public pricing page still showed the higher rates as of June 17. The company had not announced a timeline. One enterprise buyer told Free AI News they paused a $300,000 annual renewal while waiting for confirmation. OpenAI faced the same free tier squeeze as Google. ChatGPT free users began seeing ads in some regions in early June, as reported in ChatGPT free tier ads 2026.
The competitive context was broader than one cut. Anthropic replaced its flat-rate agent subsidy with a credit pool on June 15, 2026. That change did not lower token prices. Claude Opus 4.8 still cost $15.00 per million input tokens and $75.00 per million output tokens, as outlined in Anthropic ends agent subsidy June 15 credit pool replaces flat-rate access. xAI had already made Grok v9 Medium free for verified users, undercutting Google’s new output price. Stanford HAI’s 2026 AI Index ranked OpenAI, Google, and Anthropic ahead of xAI in enterprise adoption. The price war was real, but it did not benefit every user equally. Free tier limits, ads, and credit resets continued to tighten across major providers. The full free tier picture is tracked in AI free tier shifts major providers adjust pricing access June 2026.
How Do the Top Options Compare?
| Provider | Model | Input Price | Output Price | Free Tier Status |
|---|---|---|---|---|
| Gemini 2.5 Pro | $0.75 per 1M tokens | $2.50 per 1M tokens | Limited free tier, 5 RPM | |
| OpenAI | GPT-5.1 | $3.00 per 1M tokens (proposed cut to $1.80) | $12.00 per 1M tokens (proposed cut to $7.20) | Free tier now with ads |
| Anthropic | Claude Opus 4.8 | $15.00 per 1M tokens | $75.00 per 1M tokens | Credit pool replaces flat free access |
| xAI | Grok v9 Medium | $0.60 per 1M tokens | $2.40 per 1M tokens | Free for verified users |
Prices are pay-as-you-go API rates as of June 17, 2026. Free tier limits vary by region and verification status. OpenAI proposed cuts were shared with enterprise customers and may not be final.
1. Google Gemini 2.5 Pro API , Best for high-volume developers cutting token costs
Google cut input pricing 40% and output pricing 33% on June 17, 2026. The new rate of $0.75 per million input tokens undercut OpenAI’s GPT-5.1 list price by more than half. Google published the change on its official AI pricing page. The move followed earlier subscription price cuts and a tightening of free API access. Developers on the free tier saw pro model access removed for some endpoints earlier in June. For paying customers, the math changed immediately. A workload of 100 million input tokens and 10 million output tokens cost $100 under the new rates. That same workload cost $187.50 before the cut. The price cut did not remove rate limits. Google kept context caching discounts in place for Gemini 2.5 Pro. Read more in Google AI price cuts should make OpenAI Anthropic nervous.
Key strengths:
- ✅ 40% lower input cost makes Gemini 2.5 Pro one of the cheapest frontier models
- ✅ Output pricing dropped 33%, reducing long-generation costs
- ✅ Context caching still reduces repeat prompt expenses
- ✅ Official pricing page updated same day
- ❌ Free tier access to pro models remained limited
- ❌ Rate limits stayed unchanged for paid API customers
- ❌ No change to consumer app subscription limits
Who it’s for: Developers and startups that send large API token volumes and need predictable pay-as-you-go pricing.
2. OpenAI GPT-5.1 API , Best for enterprise tools awaiting a possible price drop
OpenAI had not yet cut list prices on June 17, 2026. But two enterprise account managers told customers the company was weighing matching cuts to GPT-5.1 API rates. Current list prices sat at $3.00 per million input tokens and $12.00 per million output tokens. A proposed cut would bring input pricing to $1.80 and output pricing to $7.20. OpenAI’s pricing page still showed the higher rates as of June 17. The company charged a premium for GPT-5.1 multimodal reasoning. Google’s cut made that premium harder to justify. OpenAI also faced pressure from its own free tier. In early June, ChatGPT free users began seeing ads in some regions. The company had not announced any free tier API expansion. Enterprise buyers told Free AI News they were delaying renewal decisions until OpenAI confirmed new rates. More context is available in ChatGPT pricing changes 2026.
Key strengths:
- ✅ GPT-5.1 remained a leader in reasoning tasks
- ✅ Proposed 40% cut would narrow the gap with Google
- ✅ Enterprise volume discounts may stack with list price cuts
- ✅ Existing API customers kept access to prior model versions
- ❌ Current list price was 4x Google’s new Gemini input rate
- ❌ Free tier API access remained very limited
- ❌ Proposed cuts had no public launch date
Who it’s for: Enterprise teams that already run GPT-5.1 workloads and want to wait for a confirmed price cut before committing.
3. Anthropic Claude Opus 4.8 API , Best for agent-heavy teams reworking budgets after June 15
Anthropic did not cut API prices. Instead, on June 15, 2026, it replaced its flat-rate agent subsidy with a credit pool. Claude Opus 4.8 list prices remained at $15.00 per million input tokens and $75.00 per million output tokens. That made Opus 4.8 the most expensive frontier API among major providers. The June 15 change meant free and paid users shared a monthly credit pool. Once exhausted, users had to buy more credits. Anthropic’s announcement framed the shift as transparency. Many developers saw it as a price increase for agent workloads. The credit pool also reset every five hours for free users. That reset limit was a 50 percent jump over prior Claude Code limits for paid users. But the underlying token cost stayed high. Google’s cut made Anthropic’s position more exposed. Enterprise customers told Free AI News they were testing Gemini 2.5 Pro as a fallback. See Anthropic ends agent subsidy June 15 credit pool replaces flat-rate access.
Key strengths:
- ✅ Claude Opus 4.8 still led on long-context agent tasks
- ✅ Credit pool gave users clearer spend visibility
- ✅ Five-hour reset reduced long waits for free tier
- ❌ List prices remained 20x Google’s new input rate
- ❌ Agent subsidy ended, raising costs for heavy users
- ❌ Credit pool added complexity for budget planning
Who it’s for: Developers locked into Claude’s agent workflows who can accept credit pooling and higher per-token costs.
4. xAI Grok v9 Medium , Best for free users wanting a no-cost model after paid rivals tighten
xAI opened Grok v9 Medium to free verified users in May 2026. The model cost $0.60 per million input tokens and $2.40 per million output tokens on the API. That made it cheaper than Google’s new Gemini 2.5 Pro input price but slightly cheaper on output. xAI’s pricing page listed no free API tier. The free consumer access required account verification. The move undercut paid plans at Google and OpenAI. Free users hit rate limits after a set number of prompts. xAI did not cut API prices in June. It held firm while Google cut and OpenAI weighed. The company positioned Grok v9 Medium as a developer alternative. But its integration support remained thinner than OpenAI or Google. Hugging Face hosted model weights for Grok v1 only, not v9. Stanford HAI’s 2026 AI Index ranked xAI fourth in enterprise API adoption, behind OpenAI, Google, and Anthropic. More on xAI’s free access is in Grok v9 Medium free users 2026.
Key strengths:
- ✅ Lowest input price among major API providers at $0.60
- ✅ Free verified consumer access after rivals added ads and limits
- ✅ API output price undercut Google and OpenAI
- ❌ No free API tier for developers
- ❌ Smaller tooling and integration support than OpenAI or Google
- ❌ Rate limits on free consumer access were strict
Who it’s for: Cost-sensitive developers and verified consumers who want a free or cheap frontier alternative.
Frequently Asked Questions
Did Google actually cut Gemini API prices?
Yes. On June 17, 2026, Google reduced Gemini 2.5 Pro input price 40% to $0.75 per million tokens and output price 33% to $2.50 per million tokens, according to the official Google AI pricing page.
Is OpenAI cutting GPT-5.1 API prices?
Not yet. Enterprise account managers told some customers on June 17, 2026 that OpenAI was considering a matching 40% cut to $1.80 input and $7.20 output per million tokens, but the public pricing page had not changed.
Did free tiers change?
Yes. Google had already tightened free API access to pro models, ChatGPT free users saw ads in some regions, and Anthropic replaced flat agent access with a credit pool on June 15, 2026.
What did Anthropic change on June 15?
Anthropic ended its agent subsidy and switched to a monthly credit pool that resets every five hours. Claude Opus 4.8 token prices stayed at $15 input and $75 output per million tokens.
Which provider is cheapest now?
Gemini 2.5 Pro had the cheapest frontier API rate at $0.75 input. Grok v9 Medium was lower at $0.60 input but had no free API tier and smaller tooling support.
Will this price war help consumers?
Somewhat. Paid API customers got lower Google rates, but free tier users faced more ads, tighter limits, and credit pool changes across Google, OpenAI, and Anthropic.
What Should You Remember?
- Google cut Gemini 2.5 Pro input price 40% on June 17, 2026, to $0.75 per million tokens.
- OpenAI considered matching cuts for GPT-5.1 but had not changed public list prices.
- Anthropic did not cut Claude Opus 4.8 API prices; it replaced agent subsidies with a credit pool on June 15.
- Free tiers tightened across Google, OpenAI, and Anthropic, adding ads, limits, and credit resets.
- xAI undercut Google on input pricing with Grok v9 Medium at $0.60 per million tokens but offered no free API tier.
- High-volume developers saved 46 percent on a 100M input and 10M output token Gemini workload after Google’s cut.
Free AI News is an independent editorial publication. Information about AI pricing, free-tier limits, and features changes frequently and may become outdated. Always verify current details through the vendor’s official pages. Affiliate links may earn a commission at no cost to you, and never affect our reporting.
