Is Gemini Cheaper Than Claude? API Pricing 2026
Short answer: is Gemini cheaper than Claude? At the budget end, yes, by a wide margin: Gemini 2.5 Flash-Lite costs $0.10/$0.40 per million input/output tokens against Claude Haiku 4.5’s $1/$5. In the middle, no: Gemini 3.1 Pro Preview ($2/$12) costs more per output token than Claude Sonnet 5 ($2/$10), and more again on prompts over 200K tokens. Gemini has a free tier; both give 50% off batch jobs.
TL;DR
- Cheapest tokens: Gemini Flash-Lite, about 10× cheaper than Claude Haiku 4.5.
- Pro vs Sonnet: same input price; Claude Sonnet 5 is cheaper on output.
- Prompts over 200K tokens: Claude is roughly half the price of Gemini Pro.
- Free tier: Gemini only.
- Batch jobs: 50% off on both.
All prices below come from Google’s Gemini API pricing page and Anthropic’s Claude pricing page, checked on September 26, 2026. Both change often, so check them again before you commit a budget.
Gemini vs Claude API pricing, model by model
Prices are per million tokens, standard (non-batch) rates.
| Model | Input | Output | Cached input (read) | Context window | Free tier |
|---|---|---|---|---|---|
| Gemini 2.5 Flash-Lite | $0.10 | $0.40 | $0.01 | 1M | Yes |
| Gemini 3.1 Flash-Lite | $0.25 | $1.50 | $0.025 | 1M | Yes |
| Gemini 3.8 Flash | $0.75* | $3.75* | $0.075* | 1M | Yes |
| Gemini 2.5 Pro | $1.25 ($2.50 over 200K) | $10 ($15 over 200K) | $0.125 | 1M | Yes |
| Gemini 3.1 Pro Preview | $2 ($4 over 200K) | $12 ($18 over 200K) | $0.20 | 1M | No |
| Claude Haiku 4.5 | $1 | $5 | $0.10 | 200K | No |
| Claude Sonnet 5 | $2 | $10 | $0.20 | 1M | No |
| Claude Opus 5.5 | $4 | $20 | $0.20 | 1M | No |
| Claude Fable 5.1 | $10 | $50 | $0.25 | 1M | No |
*Gemini 3.8 Flash is on launch pricing through December 31, 2026. From January 1, 2027 it doubles to $1.50 input and $7.50 output, which is more than Claude Haiku 4.5.
Three things the table doesn’t show:
- Long prompts. Gemini’s Pro models charge more once a prompt passes 200K tokens. Claude Sonnet 5, Opus 5.5 and Fable 5.1 bill the full 1M-token window at the standard rate, so a 300K-token prompt with a 2K-token answer costs about $0.62 on Sonnet 5 and about $1.24 on Gemini 3.1 Pro Preview.
- Tokens aren’t the same size. Each vendor splits text differently, and Anthropic says its models from Claude 4.7 on use a tokenizer that produces about 30% more tokens for the same text than older Claude models. Compare costs on your own prompts, not just the rate card.
- Gemini 3.1 Pro is still a preview model, and the newest Gemini Flash is 3.8. Claude’s current lineup is Fable 5.1, Opus 5.5, Sonnet 5 and Haiku 4.5.
Batch and caching discounts
| Gemini | Claude | |
|---|---|---|
| Batch API | 50% off input and output | 50% off input and output |
| Cached input | About 10% of the input price, plus an hourly storage fee ($4.50 per million tokens per hour on the Pro models) | Cache reads: $0.20 on Sonnet 5 and Opus 5.5, $0.10 on Haiku 4.5. Cache writes cost 1.25× input for a 5-minute cache or 2× for 1 hour |
| Discounts stack? | Batch and caching both apply to batch requests | Yes, batch and caching combine |
Older versions of this post said Claude had no batch option. That was wrong: Anthropic’s Batch API gives the same 50% discount as Google’s.
How do the free tiers compare?
This is where Google still has a clear advantage.
| Gemini (Google AI Studio) | Claude (Anthropic API) | |
|---|---|---|
| Free tier | Yes, on the Flash and Flash-Lite models and Gemini 2.5 Pro | No ongoing free tier |
| Not free | Gemini 3.1 Pro Preview | Everything after the starter credit |
| Trial credit | Not needed | A small amount of free credit for new users |
| Catch | Free-tier prompts may be used to improve Google’s products; lower rate limits | None, but you pay from the first real request |
For prototypes, hackathons and personal projects, Gemini’s free tier can take API costs to zero. Once a project needs higher rate limits or keeps customer data out of training, you’re comparing paid tiers anyway.
Which is better for coding?
Price per token only matters if the output is good enough. We don’t run our own benchmark, and model rankings move every few months, so check a current leaderboard such as SWE-bench and, more usefully, run your own last 20 real tasks through both. Two practical points hold regardless of rankings: Claude Sonnet 5 and Gemini 3.1 Pro Preview cost the same per input token, so for input-heavy coding work (large files, long context) the output price and the long-prompt surcharge decide the bill; and a cheaper model that needs a second attempt costs more than a pricier one that gets it right first time.
What does a model-routing strategy cost?
The cheapest setup uses both, routing each task by difficulty. The costs below assume 2,000 input and 500 output tokens per call at standard list prices, with no caching or batch discount.
| Task type | Model | Cost per 1,000 calls | vs. all on Sonnet 5 |
|---|---|---|---|
| Classification, extraction | Gemini 2.5 Flash-Lite | $0.40 | 96% less |
| Summaries, simple chat | Gemini 3.8 Flash (launch price) | $3.38 | 63% less |
| Everyday coding help | Claude Sonnet 5 | $9.00 | Baseline |
| Same task on Gemini Pro | Gemini 3.1 Pro Preview | $10.00 | 11% more |
| Hard reasoning, agentic work | Claude Opus 5.5 | $18.00 | 2× |
Sending 60% of calls to Gemini 2.5 Flash-Lite, 25% to Gemini 3.8 Flash and 15% to Claude Sonnet 5 works out at about $2.43 per 1,000 calls, 73% less than sending everything to Sonnet 5. The trade-off is routing logic to maintain and the occasional quality miss on edge cases.
How do you choose between them for a new project?
Is cost your main constraint? Start on Gemini Flash-Lite or Flash and move individual endpoints up to Claude only when quality problems show up.
Is code quality your main constraint? Default to Claude Sonnet 5 and push only high-volume, low-stakes calls to Gemini Flash.
Are your prompts huge? Above 200K tokens, Claude’s flat pricing beats Gemini Pro’s long-context rates.
Are you prototyping? Use Gemini’s free tier, then compare paid tiers once the product direction is settled.
Building a developer tool? Let users bring their own key and pick a model, so you don’t have to choose for them.
How do you track API costs across providers?
Each provider has its own dashboard: the Claude Console for Anthropic, Google AI Studio or Cloud Billing for Gemini. Check both weekly while you’re tuning routing.
If your Claude usage is mostly Claude Code on a Pro or Max plan rather than raw API calls, FavTray’s free AI Usage Tracker shows how much of your 5-hour and weekly limits you’ve used, plus your Gemini CLI quota, in the Mac menu bar; how Claude usage limits work explains the windows. For the plan side of the decision, see the Claude Code pricing breakdown and whether Claude Max is worth it. For OpenAI and DeepSeek on the same basis, see the full LLM pricing comparison and Claude vs OpenAI pricing.
The Gemini vs Claude choice comes down to tier and prompt size. For the cheapest tokens on simple, high-volume work, Gemini Flash-Lite is hard to beat. For coding and long prompts, Claude Sonnet 5 is the same price or cheaper than Gemini’s Pro model. Most teams that build seriously with LLM APIs end up using both.
Frequently Asked Questions
Is the Gemini API cheaper than the Claude API?
At the budget end, yes, by a wide margin: Gemini 2.5 Flash-Lite costs $0.10 input and $0.40 output per million tokens, against $1 and $5 for Claude Haiku 4.5. In the middle, no: Gemini 3.1 Pro Preview ($2/$12) charges more per output token than Claude Sonnet 5 ($2/$10), and much more on prompts over 200K tokens. Prices from Google's and Anthropic's pricing pages, September 26, 2026.
Is Gemini cheaper than Claude for coding?
Usually not at the quality level most coding needs. Gemini 3.1 Pro Preview and Claude Sonnet 5 both cost $2 per million input tokens, but Gemini charges $12 per million output tokens against Claude's $10, and $4/$18 once a prompt passes 200K tokens, which large codebases often do. Gemini Flash models are much cheaper for simple edits and boilerplate. Prices checked September 26, 2026.
Is Claude more expensive than Gemini for long prompts?
No, the reverse. Claude Sonnet 5, Opus 5.5 and Fable 5.1 bill the whole 1M-token window at their standard rate, while Gemini's Pro models charge more above 200K tokens. A 300K-token prompt with a 2K-token answer costs about $0.62 on Sonnet 5 and about $1.24 on Gemini 3.1 Pro Preview.
Which API has the best free tier?
Gemini. Google AI Studio has a free tier for the Flash and Flash-Lite models and Gemini 2.5 Pro (not Gemini 3.1 Pro Preview), with rate limits, and free-tier prompts may be used to improve Google's products. Anthropic gives new API users a small amount of free credit, then it's pay as you go.
How does Gemini Pro compare to Claude Sonnet on price?
Gemini 3.1 Pro Preview costs $2 input and $12 output per million tokens for prompts up to 200K tokens, and $4/$18 above that. Claude Sonnet 5 costs $2/$10 at any length up to its 1M-token window. Same input price; Claude is about 17% cheaper on output for short prompts and roughly half the price on long ones.
Does Claude have a batch API discount?
Yes. Anthropic's Batch API takes 50% off input and output tokens for asynchronous jobs, the same discount Google gives on Gemini batch requests. Both also discount cached input: Claude cache reads cost $0.20 per million tokens on Sonnet 5 and Opus 5.5; Gemini charges about 10% of the input price for cached tokens, plus an hourly storage fee.
Can you use Gemini and Claude together?
Yes. Send high-volume simple work (classification, extraction, short summaries) to a Gemini Flash or Flash-Lite model and keep Claude Sonnet 5 or Opus 5.5 for coding and multi-step reasoning. On list prices, routing 60% of calls to Gemini 2.5 Flash-Lite, 25% to Gemini 3.8 Flash and 15% to Sonnet 5 costs about 73% less than sending everything to Sonnet 5.
How do you track usage across both?
For API spend, use each provider's console: the Claude Console for Anthropic, and Google AI Studio or Cloud Billing for Gemini. If you use the subscription tools, FavTray's free AI Usage Tracker shows your Claude plan's 5-hour and weekly limits and your Gemini CLI quota in the Mac menu bar.