Together.ai SEO Issues, Traffic and Optimization Tips

Together.ai

February, 27 2024 05:42:51 AM

Passed

To Improve

Errors

0Score

Download Report Update Compare Share

Search Engine Optimization

Title Tag

Together AI
Length: 11 character(s)

Ideally, your title tag should contain between 10 and 70 characters (spaces included).
Make sure your title is explicit and contains your most important keywords.
Be sure that each page has a unique title.

Meta Description

Build gen AI models with Together AI. Benefit from the fastest and most cost-efficient tools and infra. Collaborate with our expert AI team that’s dedicated to your success.
Length: 173 character(s)

Meta descriptions contains between 100 and 300 characters (spaces included).
It allow you to influence how your web pages are described and displayed in search results.
Ensure that all of your web pages have a unique meta description that is explicit and contains your most important keywords (these appear in bold when they match part or all of the user's search query).
A good meta description acts as an organic advertisement, so use enticing messaging with a clear call to action to maximize click-through rate.

Meta Keywords

No Keywords

Meta Keywords are a specific type of meta tag that appear in the HTML code of a Web page and help tell search engines what the topic of the page is.
However, google can't use meta keywords.

Headings

<H1>	<H2>	<H3>	<H4>	<H5>	<H6>
1	66	22	3	0	0

<H1> together. </H1>
<H2> 01 Together INFERENCE </H2>
<H2> 02 Together FINE-TUNING </H2>
<H2> 03 Together gpu clusters </H2>
<H2> 04 Together custom models </H2>
<H2> THE FASTEST CLOUD FOR GEN AI. </H2>
<H2> BUILT ON LEADING AI RESEARCH. </H2>
<H2> Innovations </H2>
<H2> LLaMA-2 Chat </H2>
<H2> Mixtral Instruct </H2>
<H2> Striped Hyena Nous </H2>
<H2> Mistral Instruct </H2>
<H2> 01-AI Yi </H2>
<H2> UAE-Large v1 </H2>
<H2> Stable Diffusion XL 1.0 </H2>
<H2> M2-BERT 80M 32K Retrieval </H2>
<H2> Code Llama Instruct </H2>
<H2> RedPajama-INCITE Instruct </H2>
<H2> Vicuna v1.5 16K </H2>
<H2> Nous Capybara </H2>
<H2> Nexus Raven </H2>
<H2> Nous Hermes Llama-2 </H2>
<H2> Wizard LM </H2>
<H2> WizardCoder Python v1.0 </H2>
<H2> M2-BERT 80M 8K Retrieval </H2>
<H2> M2-BERT 80M 2K Retrieval </H2>
<H2> BGE-Large-EN v1.5 </H2>
<H2> LLaMA-2-7B-32K-Instruct </H2>
<H2> BGE-Base-EN v1.5 </H2>
<H2> Mistral </H2>
<H2> Code Llama Python </H2>
<H2> Stable Diffusion 2.1 </H2>
<H2> Vicuna v1.5 </H2>
<H2> Phind Code LLaMA v2 </H2>
<H2> Realistic Vision 3.0 </H2>
<H2> LLaMA-2 </H2>
<H2> Code Llama </H2>
<H2> LLaMA-2-32K </H2>
<H2> Chronos Hermes </H2>
<H2> Platypus2 Instruct </H2>
<H2> WizardLM v1.0 (70B) </H2>
<H2> MythoMax-L2 </H2>
<H2> Qwen-Chat </H2>
<H2> Qwen </H2>
<H2> RedPajama-INCITE Chat </H2>
<H2> RedPajama-INCITE </H2>
<H2> BERT </H2>
<H2> Sentence-BERT </H2>
<H2> GPT-JT-Moderation </H2>
<H2> GPT-JT </H2>
<H2> GPT-NeoXT-Chat-Base </H2>
<H2> LLaMA </H2>
<H2> Falcon Instruct </H2>
<H2> Falcon </H2>
<H2> Openjourney v4 </H2>
<H2> Pythia-Chat-Base </H2>
<H2> Analog Diffusion </H2>
<H2> Together Inference </H2>
<H2> Together Fine-tuning </H2>
<H2> Need a cluster of H100s on Infiniband today? </H2>
<H2> Together GPU Clusters </H2>
<H2> Together Custom Models </H2>
<H2> We believe the future of AI is </H2>
<H2> Open Source </H2>
<H2> Build, deploy, and scale. All in a single platform. </H2>
<H2> Customer Stories </H2>
<H2> Subscribe to newsletter </H2>
<H3> Cocktail SGD </H3>
<H3> FlashAttention 2 </H3>
<H3> Sub-quadratic model architectures </H3>
<H3> RedPajama </H3>
<H3> speed relative to tgi, vllm orother inference services </H3>
<H3> llama-2 70b </H3>
<H3> cost relative to gpt-3.5-turbo </H3>
<H3> 01 </H3>
<H3> 02 </H3>
<H3> 03 </H3>
<H3> training </H3>
<H3> network c0mmunication </H3>
<H3> relative to aws </H3>
<H3> 01 Transparency </H3>
<H3> 02 no lock-in </H3>
<H3> 03 Privacy </H3>
<H3> Build </H3>
<H3> Deploy </H3>
<H3> Scale </H3>
<H3> Pika creates the next gen text-to-video models on Together GPU Clusters </H3>
<H3> Nexusflow uses Together GPU Clusters to build cybersecurity models </H3>
<H3> Arcee builds domain adaptive language models with Together Custom Models </H3>
<H4> High end clusters without long-term commits </H4>
<H4> Clusters from 16 GPUs to 2048 GPUs </H4>
<H4> Snappy setup. Blazing fast training. </H4>

Show More

Show Less

Use your keywords in the headings and make sure the first level (H1) includes your most important keywords. Never duplicate your title tag content in your header tag.
While it is important to ensure every page has an H1 tag, never include more than one per page. Instead, use multiple H2 - H6 tags.

Google Preview

Alt Attribute

We found 9 images on this web page

7 ALT attributes are empty or missing.

https://assets-global.website-files.com/650c3b59079d92475f37b68f/653bf70f91e3154115d3df0b_pikalabs.png
https://assets-global.website-files.com/650c3b59079d92475f37b68f/656671bcb5eedcfea92b1d5a_Pika_Avatar.png
https://assets-global.website-files.com/650c3b59079d92475f37b68f/654109a9abf6ff146d4c27c1_nexusflow-white-300.png
https://assets-global.website-files.com/650c3b59079d92475f37b68f/65458ff56464c3cb6a12622b_Screenshot%202023-11-03%20at%205.27.27%E2%80%AFPM.png
https://assets-global.website-files.com/650c3b59079d92475f37b68f/654141dfbfd06a9a1b6bf179_arcee-white-300.png
https://assets-global.website-files.com/650c3b59079d92475f37b68f/6543d9c2fa14007cbaf19814_dalm.jpeg
https://assets-global.website-files.com/64f6f2c0e3f4c5a91c1e823a/6500732503885fd3e7e06d70_logo-dark.svg

Show More

Show Less

Alternative text is used to describe images to give the search engine crawlers (and the visually impaired).
Also, more information to help them understand images, which can help them to appear in Google Images search results.

Keywords Cloud

model44
models23
together21
language15
large12
data12
llama11
been10
trained10
clusters9
fine-tuned9
code9
based9
like9
published8

This Keyword Cloud provides an insight into the frequency of keyword usage within the page.
It's important to carry out keyword research to get an understanding of the keywords that your audience is using. There are a number of keyword research tools available online to help you choose which keywords to target.

Keyword Consistency

Text/HTML Ratio

HTML to Text Ratio is: 12.76%

Text content size	24401 bytes
Total HTML size	191159 bytes

Code to text ratio represents the percentage of actual text on a web page compared to the percentage of HTML code, and it is used by search engines to calculate the relevancy of a web page.
A higher code to text ratio will increase your chances of getting a better rank in search engine results.

GZIP compression

WWW Resolve

IP Canonicalization

No your domain IP 99.83.190.102 does not redirect to together.ai

To check this for your website, enter your IP address in the browser and see if your site loads with the IP address.
Ideally, the IP should redirect to your website's URL or to a page from your website hosting provider.
If it does not redirect, you should do an htaccess 301 redirect to make sure the IP does not get indexed.

XML Sitemap

Robots.txt

URL Rewrite

Good, all URLs look clean and friendly

Your site's URLs contain unnecessary elements that make them look complicated.
A URL must be easy to read and remember for users. Search engines need URLs to be clean and include your page's most important keywords.
Clean URLs are also useful when shared on social media as they explain the page's content.

Underscores in the URLs

Great, you are not using underscores (these_are_underscores) in your URLs

Great, you are not using ?underscores (these_are_underscores) in your URLs.
While Google treats hyphens as word separators, it does not for underscores.

Embedded Objects

Iframe

Perfect, no Iframe content has been detected on this page

Frames can cause problems on your web page because search engines will not crawl or index the content within them.
Avoid frames whenever possible and use a NoFrames tag if you must use them.

Domain Registration

WHOIS Data

Indexed Pages

Indexed pages in search engines

325 Page(s)

This is the number of pages that we have discovered on your website.
A low number can indicate that bots are unable to discover your webpages, which is a common cause of a bad site architecture & internal linking, or you're unknowingly preventing bots and search engines from crawling & indexing your pages.

Backlinks Counter

Number of backlinks to your website

0 Backlink(s)

Backlinks are links that point to your website from other websites. They are like letters of recommendation for your site.
Since this factor is crucial to SEO, you should have a strategy to improve the quantity and quality of backlinks.

Usability

URL

http://together.ai
Length: 8 characters

Keep your URLs short and avoid long domain names when possible.
A descriptive URL is better recognized by search engines.
A user should be able to look at the address bar and make an accurate guess about the content of the page before reaching it (e.g., http://www.mysite.com/en/products).

Favicon

Great, your website has a favicon.

Favicons improve a brand's visibility.
As a favicon is especially important for users bookmarking your website, make sure it is consistent with your brand.

Custom 404 Page

Great, your website has a custom 404 error page.

When a visitor encounters a 404 File Not Found error on your site, you're on the verge of losing the visitor that you've worked so hard to obtain through the search engines and third party links.
Creating your custom 404 error page allows you to minimize the number of visitors lost that way.

Page Size

187 KB (World Wide Web average is 320 Kb)

Two of the main reasons for an increase in page size are images and JavaScript files.
Page size affects the speed of your website; try to keep your page size below 2 Mb.
Tip: Use images with a small size and optimize their download with gzip.

Load Time

PageSpeed Insights (Desktop)

Language

Good, you have declared your language
Declared Language: English

Make sure your declared language is the same as the language detected by Google
Also, define the language of the content in each page's HTML code.

Domain Availability

Typo Availability

Domains (TLD)	Status
rogether.ai	Query Failed
fogether.ai	Query Failed
gogether.ai	Query Failed
hogether.ai	Query Failed
yogether.ai	Query Failed

Email Privacy

Safe Browsing

The website is not blacklisted and looks safe to use.

Safe Browsing to identify unsafe websites and notify users and webmasters so they can protect themselves from harm.

Mobile

Mobile Friendliness

Mobile View

The number of people using the Mobile Web is huge; over 75 percent of consumers have access to smartphones. ??
Your website should look nice on the most popular mobile devices.
Tip: Use an analytics tool to track mobile usage of your website.

Mobile Compatibility

PageSpeed Insights (Mobile)

Technologies

Server IP

Speed Tips

Analytics

W3C Validity

W3C not validated

W3Cis a consortium that sets web standards.
Using valid markup that contains no errors is important because syntax errors can make your page difficult for search engines to index. Run the W3C validation service whenever changes are made to your website's code.

Doc Type

Encoding

Great, language/character encoding is specified: UTF-8

Specifying language/character encoding can prevent problems with the rendering of special characters.

Link Analysis

In-Page Links

We found a total of 86 links including both internal & external links of your site

Anchor	Type	Follow
Announcing our $102.5M Series A	Internal Links	Dofollow
http://together.ai/	Internal Links	Dofollow
Products	Internal Links	Dofollow
Together Inference	Internal Links	Dofollow
Together Fine-tuning	Internal Links	Dofollow
Together Custom Models	Internal Links	Dofollow
Together GPU Clusters	Internal Links	Dofollow
Solutions	Internal Links	Dofollow
What we offer	Internal Links	Dofollow
Customer stories	Internal Links	Dofollow
Why open-source	Internal Links	Dofollow
Industries & use cases	Internal Links	Dofollow
Research	Internal Links	Dofollow
Blog	Internal Links	Dofollow
About	Internal Links	Dofollow
Values	Internal Links	Dofollow
Careers	Internal Links	Dofollow
Team	Internal Links	Dofollow
Pricing	Internal Links	Dofollow
Contact	Internal Links	Dofollow
Cocktail SGD With Cocktail SGD, we’ve addressed a key hindrance to training generative AI models in a distributed environment: networking overhead. Cocktail SGD is a set of optimizations that reduces network overhead by up to 117x.Read more	Internal Links	Dofollow
FlashAttention 2 FlashAttention is now broadly used by all Transformer models. It accelerates training and fine-tuning of LLMs by up to 9x and achieves an unprecedented 72% model FLOPs utilization for training on NVIDIA A100s.Read more	Internal Links	Dofollow
Sub-quadratic model architectures In close collaboration with Hazy Research, we’re working on the next core architecture for generative AI models that will provide even faster performance with longer context. Our research published in this area includes Striped Hyena, Monarch Mixer, and FlashConv.Read more	Internal Links	Dofollow
RedPajama Our RedPajama project enables leading generative AI models to be available as fully open-source. The RedPajama models have been downloaded millions of times, and the RedPajama dataset has been used to create over 500 leading models.Read more	Internal Links	Dofollow
Reserve your cluster today	Internal Links	Dofollow
http://together.ai/solutions#pika-labs	Internal Links	Dofollow
http://together.ai/solutions#nexus-flow	Internal Links	Dofollow
http://together.ai/solutions#arcee	Internal Links	Dofollow
here	Internal Links	Dofollow
Llama-2-13B on Together Inference	Internal Links	Dofollow
this research paper.	Internal Links	Dofollow
Privacy policy	Internal Links	Dofollow
Terms of service	Internal Links	Dofollow
Docs	External Links	Dofollow
Get Started	External Links	Dofollow
Start building now	External Links	Dofollow
Docs	External Links	Dofollow
ChatLlama 2-chat leverages publicly available instruction datasets and over 1 million human annotations. Available in three sizes: 7B, 13B and 70B parametersLLaMA-2 ChatVIEW MORE	External Links	Dofollow
ChatThe Mixtral-8x7B Large Language Model (LLM) is a pretrained generative Sparse Mixture of Experts.Mixtral InstructVIEW MORE	External Links	Dofollow
ChatA hybrid architecture composed of multi-head, grouped-query attention and gated convolutions arranged in Hyena blocks, different from traditional decoder-only TransformersStriped Hyena NousVIEW MORE	External Links	Dofollow
Chatinstruct fine-tuned version of Mistral-7B-v0.1Mistral InstructVIEW MORE	External Links	Dofollow
LanguageThe Yi series models are large language models trained from scratch by developers at 01.AI01-AI YiVIEW MORE	External Links	Dofollow
EmbeddingsAn universal English sentence embedding model by WhereIsAI. Its embedding dimension is 1024, it takes up to 512 context length.UAE-Large v1VIEW MORE	External Links	Dofollow
ImageA text-to-image generative AI model that excels at creating 1024x1024 images.Stable Diffusion XL 1.0VIEW MORE	External Links	Dofollow
ChatCode Llama is a family of large language models for code based on Llama 2 providing infilling capabilities, support for large input contexts, and zero-shot instruction following ability for programming tasks.Code Llama InstructVIEW MORE	External Links	Dofollow
LanguageDesigned for few-shot prompts, fine-tuned over the RedPajama-INCITE-Base-7B-v1 base model.RedPajama-INCITE InstructVIEW MORE	External Links	Dofollow
ChatVicuna is a chat assistant trained by fine-tuning Llama 2 on user-shared conversations collected from ShareGPT.Vicuna v1.5 16KVIEW MORE	External Links	Dofollow
ChatFirst Nous collection of dataset and models made by fine-tuning mostly on data created by Nous in-houseNous CapybaraVIEW MORE	External Links	Dofollow
ChatNous-Hermes-Llama2-13b is a state-of-the-art language model fine-tuned on over 300,000 instructions.Nous Hermes Llama-2VIEW MORE	External Links	Dofollow
ChatThis model achieves a substantial and comprehensive improvement on coding, mathematical reasoning and open-domain conversation capacitiesWizard LMVIEW MORE	External Links	Dofollow
CodeThis model empowers Code LLMs with complex instruction fine-tuning, by adapting the Evol-Instruct method to the domain of code.WizardCoder Python v1.0VIEW MORE	External Links	Dofollow
ChatExtending LLaMA-2 to 32K context, built with Meta's Position Interpolation and Together AI's data recipe and system optimizations, instruction tuned by TogetherLLaMA-2-7B-32K-InstructVIEW MORE	External Links	Dofollow
Language7.3B parameter model that outperforms Llama 2 13B on all benchmarks, approaches CodeLlama 7B performance on code, Uses Grouped-query attention (GQA) for faster inference and Sliding Window Attention (SWA) to handle longer sequences at smaller costMistralVIEW MORE	External Links	Dofollow
CodeCode Llama is a family of large language models for code based on Llama 2 providing infilling capabilities, support for large input contexts, and zero-shot instruction following ability for programming tasks.Code Llama PythonVIEW MORE	External Links	Dofollow
ImageLatent text-to-image diffusion model capable of generating photo-realistic images given any text input.Stable Diffusion 2.1VIEW MORE	External Links	Dofollow
ChatVicuna is a chat assistant trained by fine-tuning Llama 2 on user-shared conversations collected from ShareGPT.Vicuna v1.5VIEW MORE	External Links	Dofollow
LanguagePhind-CodeLlama-34B-v1 trained on additional 1.5B tokens high-quality programming-related data proficient in Python, C/C++, TypeScript, Java, and more.Phind Code LLaMA v2VIEW MORE	External Links	Dofollow
ImageFine-tune version of Stable Diffusion focused on photorealism.Realistic Vision 3.0VIEW MORE	External Links	Dofollow
LanguageLanguage model trained on 2 trillion tokens with double the context length of Llama 1. Available in three sizes: 7B, 13B and 70B parametersLLaMA-2VIEW MORE	External Links	Dofollow
CodeCode Llama is a family of large language models for code based on Llama 2 providing infilling capabilities, support for large input contexts, and zero-shot instruction following ability for programming tasks.Code LlamaVIEW MORE	External Links	Dofollow
LanguageExtending LLaMA-2 to 32K context, built with Meta's Position Interpolation and Together AI's data recipe and system optimizations.LLaMA-2-32KVIEW MORE	External Links	Dofollow
ChatThis model is a 75/25 merge of Chronos (13B) and Nous Hermes (13B) models resulting in having a great ability to produce evocative storywriting and follow a narrative.Chronos HermesVIEW MORE	External Links	Dofollow
ChatAn instruction fine-tuned LLaMA-2 (70B) model by merging Platypus2 (70B) by garage-bAInd and LLaMA-2 Instruct v2 (70B) by upstage.Platypus2 InstructVIEW MORE	External Links	Dofollow
LanguageThis model achieves a substantial and comprehensive improvement on coding, mathematical reasoning and open-domain conversation capacities.WizardLM v1.0 (70B)VIEW MORE	External Links	Dofollow
ChatMythoLogic-L2 and Huginn merge using a highly experimental tensor type merge technique. The main difference with MythoMix is that I allowed more of Huginn to intermingle with the single tensors located at the front and end of a modelMythoMax-L2VIEW MORE	External Links	Dofollow
Chat7B-parameter version of the large language model series, Qwen (abbr. Tongyi Qianwen), proposed by Aibaba Cloud. Qwen-7B-Chat is a large-model-based AI assistant, which is trained with alignment techniques.Qwen-ChatVIEW MORE	External Links	Dofollow
Language7B-parameter version of the large language model series, Qwen (abbr. Tongyi Qianwen), proposed by Aibaba Cloud. Qwen-7B is a Transformer-based large language model, which is pretrained on a large volume of data, including web texts, books, codes, etc.QwenVIEW MORE	External Links	Dofollow
ChatChat model fine-tuned using data from Dolly 2.0 and Open Assistant over the RedPajama-INCITE-Base-7B-v1 base model.RedPajama-INCITE ChatVIEW MORE	External Links	Dofollow
LanguageBase model that aims to replicate the LLaMA recipe as closely as possible (blog post).RedPajama-INCITEVIEW MORE	External Links	Dofollow
LanguageThis model can be used to moderate other chatbot models. Built using GPT-JT model fine-tuned on Ontocord.ai's OIG-moderation dataset v0.1.GPT-JT-ModerationVIEW MORE	External Links	Dofollow
LanguageFork of GPT-J instruction tuned to excel at few-shot prompts (blog post).GPT-JTVIEW MORE	External Links	Dofollow
ChatChat model fine-tuned from EleutherAI’s GPT-NeoX with over 40 million instructions on carbon reduced compute.GPT-NeoXT-Chat-BaseVIEW MORE	External Links	Dofollow
LanguageAn auto-regressive language model, based on the transformer architecture. The model comes in different sizes: 7B, 13B, 33B and 65B parameters.LLaMAVIEW MORE	External Links	Dofollow
ChatFalcon-40B-Instruct is a causal decoder-only model built by TII based on Falcon-40B and finetuned on a mixture of Baize.Falcon InstructVIEW MORE	External Links	Dofollow
LanguageFalcon-40B is a causal decoder-only model built by TII and trained on 1,000B tokens of RefinedWeb enhanced with curated corpora.FalconVIEW MORE	External Links	Dofollow
ImageAn open source Stable Diffusion model fine tuned model on Midjourney images.Openjourney v4VIEW MORE	External Links	Dofollow
ChatChat model based on EleutherAI’s Pythia-7B model, and is fine-tuned with data focusing on dialog-style interactions.Pythia-Chat-BaseVIEW MORE	External Links	Dofollow
ImageDreambooth model trained on a diverse set of analog photographs to provide an analog film effect.Analog DiffusionVIEW MORE	External Links	Dofollow
Mosaic ML blog	External Links	Dofollow
Open AI GPT-3.5-Turbo	External Links	Dofollow
Source	External Links	Dofollow
AWS Capacity Blocks	External Links	Dofollow
AWS p5.48xlarge	External Links	Dofollow
https://discord.gg/9Rk6sSeWEG	External Links	Dofollow
http://twitter.com/togethercompute	External Links	Dofollow
https://www.linkedin.com/company/togethercomputer	External Links	Dofollow

Show More

Show Less

While there's no exact limit to the number of links you should include on a page, best practice is to avoid exceeding 200 links.
Links pass value from one page to another, but the amount of value that can be passed is split between all of the links on a page. This means that adding unnecessary links will dilute the potential value attributed to your other links.
Using the Nofollow attribute prevents value from being passed to the linking page, but it's worth noting that these links are still taken into account when calculating the value that is passed through each link, so Nofollow links can also dilute pagerank.

Together.ai

Search Engine Optimization

Title Tag

Meta Description

Meta Keywords

Headings

Google Preview

Alt Attribute

Keywords Cloud

Keyword Consistency

Text/HTML Ratio

GZIP compression

WWW Resolve

IP Canonicalization

XML Sitemap

Robots.txt

URL Rewrite

Underscores in the URLs

Embedded Objects

Iframe

Domain Registration

WHOIS Data

Indexed Pages

Backlinks Counter

Usability

URL

Favicon

Custom 404 Page

Page Size

Load Time

PageSpeed Insights (Desktop)

Language

Domain Availability

Typo Availability

Email Privacy

Safe Browsing

Mobile

Mobile Friendliness

Mobile View

Mobile Compatibility

PageSpeed Insights (Mobile)

Technologies

Server IP

Speed Tips

Analytics

W3C Validity

Doc Type

Encoding

Social

Social Data

Visitors

Estimated Worth

Traffic Rank

Visitors Localization

Link Analysis

In-Page Links

Broken Links

Try New Site