To solve this inconvenience I spent a few hours making pricepertoken.com which has the latest model's up-to-date prices all in one place.
Thinking about adding image models too especially since you have multiple options (fal, replicate) to use the same model and the prices are not always the same.
We have solved this problem by working with the providers to implement a prices and models API that we scrape, which is how we keep our marketplace up to date. It's been a journey; a year ago it was all happening through conversations in shared Slack channels!
The pricing landscape has become more complex as providers have introduced e.g. different prices for tokens depending on prompt length, caching, etc.
I do believe the right lens on this is actually the price per token by endpoint, not by model; there are fast/slow versions, thinking/non-thinking, etc. that can sometimes also vary by price.
The point of this comment is not to self promote, but we have put a huge amount of work into figuring all of this out, and have it all publicly available on OpenRouter (admittedly not in such a compact, pricing-focused format though!)
Does OpenRouter offer a similar way to select any model in a user friendly way? Meaning, can I add some budget, hook it up to a desktop application and then just have a convenient selection of models that I can start using?
And also, which model would you use to not break the bank? Currently I am mostly using Gemini 2.5 Pro but perhaps there is already a cheaper and better option.
[1] https://ai.google.dev/gemini-api/docs/pricing#gemini-2.5-fla...
edit: my bad I was wrong shouldnt have responded like this
Your website reports 0.30$ for input and that wouldn't make any sense as it would be priced the same as the bigger Flash model.
Put a really really bad taste in my mouth.
- for other models there are providers that serve the same model with different prices
- each provider optimizes for different parameters: speed, cost, etc.
- the same model can still be different quantizations
- some providers offer batch pricing (e.g., Grok API does not)
And there are plenty of other parameters to filter over- thinking vs. non-thinking, multi-modal or not, etc. not to even mention benchmarks ranking.
https://artificialanalysis.ai gives a blended cost number which helps with sorting a bit, but a blended cost model for input/output costs are going to change depending on what you're doing.
I'm still holding my breath for a site that has a really nice comparison UI.
Someone please build it!
We have a simple model comparison tool that is not-at-all-obvious to find on the website, but hopefully can help somewhat. E.g.
https://openrouter.ai/compare/qwen/qwen3-coder/moonshotai/ki...
I want model:
1) with audio input
2) minimum 50 tps speed
3) max price less than $1 input and less than $3 output
4) need to support english only or need to support polish etc.
sort by WER or some benchmark, dispaly charts etc.
edit:
extra bonus if I can tell how big typical my prompt will be like 20 seconds audio ant it will figure out how many tokens it will be because e.g. gemini 2.0 flash hide it very deep that its supposed to be 32 tokens per 1 second. Same hard to find how many tokens is for image input sometimes. Would be good where I can also attach some text or sample input for a query to do the calculation
Ideally write such prompt and it show me results or map this prompt to SQL that executes in your data so I can tweak SQL query on website. It doesn't have to be SQL can be some other simple but deterministic query language.
- An image will take 10x token on gpt-4o-mini vs gpt-4.
- On gemini 2.5 pro output token are token except if you are using structure output, then all character are count as a token each for billing.
- ...
Having the price per token is nice, but what is really needed is to know how much a given query / answer will cost you, as not all token are equals.
One other idea - for people spending $20 or $200/month for AI coding tools, a monitoring service that tracks and alerts on detected pricing changes could be something worth paying for. I'd definitely subscribe at $5/month for something like that, and I'd consider paying more, possibly even talking work into paying $20 or $30 per month.
Can you elaborate this? I don’t quite understand the difference.
- Off-peak pricing by DeepSeek
- Batch pricing by OpenAI and Anthropic
- Context window differentiated pricing by Google and Grok
- Thinking vs non-thinking token pricing by Qwen
- Input token tiered pricing by Qwen coder
I originally posted here: https://x.com/paradite_/status/1947932450212221427
Are there any tutorials you can recommend for somebody interested in getting something running locally?
You can, almost, convert the number of nodes to gb of memory needed. For example, Deepseek-r1:7b needs about 7gb of memory to run locally.
Context window matters, the more context you need, the more memory you'll need.
If you are looking for AI devices at $2500, you'll probably want something like this [1]. A unified memory architecture (which will mean LPDDR5) will give you the most memory for the least amount of money to play with AI models.
[1] https://frame.work/products/desktop-diy-amd-aimax300/configu...
When local models don’t cut it, I like Gemini 2.5 flash/pro and gemini-cli.
There are a lot of good options for commercial APIs and for running local models. I suggest choosing a good local and a good commercial API, and spend more time building things than frequently trying to evaluate all the options.
It's been a while since I checked out Mini prices. Today, $2400 buys an M4 Pro with all the cores, 64GB RAM, and 1TB storage. That's pleasantly surprising...
there are no out the box solutions to run a fleet of models simultaneously or containerized either
so the closed source solutions in the cloud are light years ahead and its been this way for 15 months now, no signs of stopping
would probably need multiple models running in distinct containers, with another process coordinating them
Openrouter is a good alternative. Added bonus that you can also see where the open models come in, and can make an educated guess on the true cost / size of a model, and how likely it is it's currently subsidised.
A limitation though, at least the last time I checked, is that you only get a single provider returned per model. That’s fine for the major commercial models that all have the same pricing on each provider, but makes it hard to rely on for open source models, which tend to have many providers offering them at different price points (sometimes very different price points—like 5x or 10x difference).
https://github.com/tekacs/llm-pricing
llm-pricing
Model | Input | Output | Cache Read | Cache Write
------------------------------------------+-------+--------+------------+------------
anthropic/claude-opus-4 | 15.00 | 75.00 | 1.50 | 18.75
anthropic/claude-sonnet-4 | 3.00 | 15.00 | 0.30 | 3.75
google/gemini-2.5-pro | 1.25 | 10.00 | N/A | N/A
x-ai/grok-4 | 3.00 | 15.00 | 0.75 | N/A
openai/gpt-4o | 2.50 | 10.00 | N/A | N/A
...
llm-pricing calc 10000 200 -c 9500 opus-4 4.1
Cost calculation: 10000 input + 200 output (9500 cached, 5m TTL)
Model | Input | Output | Cache Read | Cache Write | Total
---------------------------+-----------+-----------+------------+-------------+----------
anthropic/claude-opus-4 | $0.007500 | $0.015000 | $0.014250 | $0.178125 | $0.214875
openai/gpt-4.1 | $0.001000 | $0.001600 | $0.004750 | $0.000000 | $0.007350
openai/gpt-4.1-mini | $0.000200 | $0.000320 | $0.000950 | $0.000000 | $0.001470
openai/gpt-4.1-nano | $0.000050 | $0.000080 | $0.000237 | $0.000000 | $0.000367
thudm/glm-4.1v-9b-thinking | $0.000018 | $0.000028 | $0.000333 | $0.000000 | $0.000378
llm-pricing opus-4 -v
=== ANTHROPIC ===
Model: anthropic/claude-opus-4
Name: Anthropic: Claude Opus 4
Description: Claude Opus 4 is benchmarked as the world's best coding model, at time of release,
bringing sustained performance on complex, long-running tasks and agent workflows. It sets new
benchmarks in software engineering, achieving leading results on SWE-bench (72.5%) and
Terminal-bench (43.2%).
Pricing:
Input: $15.00 per 1M tokens
Output: $75.00 per 1M tokens
Cache Read: $1.50 per 1M tokens
Cache Write: $18.75 per 1M tokens
Per Request: $0
Image: $0.024
Context Length: 200000 tokens
Modality: text+image->text
Tokenizer: Claude
Max Completion Tokens: 32000
Moderated: trueThe UI, however, is really clean and straight to the point. I like the interface, but miss the content
Mistral, Llama, Kimi, Qwen…?
Not a site if value unless it contains whole of the market.
[0] https://developers.googleblog.com/en/gemini-2-5-models-now-s...
What tools / experiments out there exist to exercise these cheaper models to output more tokens / use more CoT tokens to achieve the quality of more expensive models?
eg, Gemini 2.5 flash / pro ratio is 1 1/3 for input, 1/8 for output... Surely there's a way to ask Flash to critique it's work more thoroughly to get to Pro level performance and still save money?
We have uploaded mostly the openrouter api models, but trying to do it in a useful way to personalize calculation and comparison. If someone would like to test or have a demo, we will be glad for any feedback.
Here's the current version:
const pricesPerMillion = {
"o1-2024-12-17": { input: 15.00, output: 60.00 },
"o1-mini-2024-09-12": { input: 1.10, output: 4.40 },
"o3-mini-2025-01-31": { input: 1.10, output: 4.40 },
"gpt-4.5-preview-2025-02-27": { input: 75.00, output: 150.00 },
"gpt-4o": { input: 5.00, output: 15.00 },
"gpt-4o-2024-08-06": { input: 2.50, output: 10.00 },
"gpt-4o-2024-05-13": { input: 5.00, output: 15.00 },
"gpt-4o-mini": { input: 0.15, output: 0.60 },
"gpt-4o-mini-2024-07-18": { input: 0.15, output: 0.60 },
"gpt-4-0613": { input: 30.00, output: 60.00 },
"gpt-4-turbo-2024-04-09": { input: 10.00, output: 30.00 },
"gpt-3.5-turbo": { input: 0.003, output: 0.006 },
"gpt-4.1": { input: 2.00, output: 8.00 },
"gpt-4.1-2025-04-14": { input: 2.00, output: 8.00 },
"gpt-4.1-mini": { input: 0.40, output: 1.60 },
"gpt-4.1-mini-2025-04-14": { input: 0.40, output: 1.60 },
"gpt-4.1-nano": { input: 0.10, output: 0.40 },
"gpt-4.1-nano-2025-04-14": { input: 0.10, output: 0.40 },
"gpt-4o-audio-preview-2024-12-17": { input: 2.50, output: 10.00 },
"gpt-4o-realtime-preview-2024-12-17": { input: 5.00, output: 20.00 },
"gpt-4o-mini-audio-preview-2024-12-17": { input: 0.15, output: 0.60 },
"gpt-4o-mini-realtime-preview-2024-12-17": { input: 0.60, output: 2.40 },
"o1-pro-2025-03-19": { input: 150.00, output: 600.00 },
"o3-pro-2025-06-10": { input: 20.00, output: 80.00 },
"o3-2025-04-16": { input: 2.00, output: 8.00 },
"o4-mini-2025-04-16": { input: 1.10, output: 4.40 },
"codex-mini-latest": { input: 1.50, output: 6.00 },
"gpt-4o-mini-search-preview-2025-03-11": { input: 0.15, output: 0.60 },
"gpt-4o-search-preview-2025-03-11": { input: 2.50, output: 10.00 },
"computer-use-preview-2025-03-11": { input: 3.00, output: 12.00 }
};
1. Pull some large tech company’s open source’ tools’ JS file 2. Extract an internal JSON blob that contains otherwise difficult information 3. Parse it and use what I need from within it for my tool
Just now I was testing the new Qwen3-thinking model. I've run the same prompt five times. The costs I got, sorted: 0.0143, 0.0288, 0.0321, 0.0389, 0.048 . And this is for single model.
Also, in my experience, sonnet-4 is cheaper than gemini-2.5-pro, despite token costs being higher.
You should also aggregate prices from Vertex and AWS Bedrock .
- Filter by model "power" or price class. I want compare the mini models, the medium models, etc.
- I'd like to see a "blended" cost which does 80% input + 20% output, so I can quickly compare the overall cost.
Great work on this!
Should I use copilot pro in agent mode with sonnet 4, or is it cheaper to use claude with sonnet 4 directly?
I am newly to this hobby, but would like to know more about what experienced person things and do.
https://claude.ai/share/20b36bd3-d817-4228-bc33-aa7c4910bc2b (the preview seems to only work in Chrome, for Firefox you have to download the html).
Plus maybe half an hour to verify and correct the prices and another few minutes for domain and hosting.
The author posted it himself, so why not spend an hour or two more and have a decent list with at least half a dozen providers and 100 models? In its current state it is just a mockup.
It is since 3 hours on the top of the front page, if the author had added one per minute to the .json, then it would now be 200 models.
I plan on adding cache prices and making the list more comprehensive