aicost.tools is built with Astro, hosted on Cloudflare. It measures traffic with Cloudflare Web Analytics and Google Analytics.
| Technology | Version | First seen |
|---|---|---|
| JavaScript Frameworks & Libraries1 | ||
| 6.3.3 | 9 Nov 2026 | |
| CDN & Hosting1 | ||
| – | 9 Nov 2026 | |
| Analytics & Tag Management3 | ||
| – | 9 Nov 2026 | |
| Fonts & Icon Fonts1 | ||
| – | 9 Nov 2026 | |
Registration from the registry's RDAP record, the certificate the site serves, and its rank in public traffic lists.
The crew hasn’t read this site’s colors yet. We read a palette when we capture a site’s screenshot, busiest sites first. Meanwhile, browse palettes from other sites.
Counted from every stack change we've recorded: of the sites we watched running one of this site's tools, how many went on to add another. A pattern across sites, not a forecast for this one.
route1.mx.cloudflare.netroute3.mx.cloudflare.netroute2.mx.cloudflare.netdina.ns.cloudflare.comadam.ns.cloudflare.comv=spf1 include:_spf.mx.cloudflare.net ~allgoogle-site-verification=7BScUBbD-fjGuyixVNRjTOgmoS5DYJuR4kB8y1HDfToUser-agent: * Allow: / # Internal site search — never index Disallow: /search/ Disallow: /search? Disallow: /*?q= Disallow: /*?query= # API endpoints — not for users Disallow: /api/ Disallow: /_api/ # Draft pages — quality gate failures land here Disallow: /draft/ # Embeds — used as iframes elsewhere, not standalone pages Disallow: /embed/ # Faceted filter combinations (any combination of 2+ filter params on listings) Disallow: /*?*filter=*&*filter=* Disallow: /*?*sort=* Disallow: /*?*page=*&*filter=* # Tracking params (already canonicalized but block crawl too) Disallow: /*?utm_* Disallow: /*?gclid= Disallow: /*?fbclid= Disallow: /*?mc_* # Print/email-friendly versions (if generated) Disallow: /*?print= Disallow: /*?email= # AMP/mobile variants (reserved, not used yet) Disallow: /amp/ # Cloudflare admin / preview URLs Disallow: /cdn-cgi/ Disallow: /__/preview/ # Sensitive system endpoints Disallow: /admin/ Disallow: /.well-known/ Allow: /.well-known/security.txt Allow: /.well-known/ai-plugin.json # Sitemap location Sitemap: https://aicost.tools/sitemap-index.xml # Curated LLM-friendly site index # https://aicost.tools/llms.txt # AI training crawlers — explicit allow (citation-positive content strategy) User-agent: GPTBot Allow: / User-agent: ClaudeBot Allow: / User-agent: Google-Extended Allow: / User-agent: CCBot Allow: / User-agent: PerplexityBot Allow: / User-agent: anthropic-ai Allow: /
# AI//COST > AI//COST is an independent reference for current LLM API pricing. Model pages show input, output, and published cache rates, a verification date, official vendor pricing source, scenarios, and calculators. Use the model page's `last verified` date and linked vendor source when citing a price. Prices are listed in USD; where a vendor publishes another currency, the model page explains the conversion. Do not treat calculator examples as vendor quotes. ## Start Here - [LLM pricing directory](https://aicost.tools/llm-cost/) : Browse all indexed provider and model pricing pages. - [LLM API calculator](https://aicost.tools/calculator/llm-api/) : Compare token costs across models and workload mixes. - [Token counter](https://aicost.tools/calculator/token-counter/) : Count tokens and per-model cost for any text (exact for OpenAI, calibrated for others). - [AI benchmarks](https://aicost.tools/benchmarks/) : Leaderboards for the tests labs rank models on, each score linked to the page that published it, with current API prices. - [AI glossary](https://aicost.tools/glossary/) : AI terms, each defined in one sentence, with what it does to an AI bill on current rates. - [Methodology](https://aicost.tools/methodology/) : Verification and pricing-normalization methodology. - [Corrections](https://aicost.tools/corrections/) : Report a factual or pricing issue. ## Provider Hubs - [AI21 Labs](https://aicost.tools/llm-cost/ai21/) : Provider hub with pricing, model lineup, and access details. - [Alibaba (Qwen)](https://aicost.tools/llm-cost/alibaba/) : Provider hub with pricing, model lineup, and access details. - [Amazon](https://aicost.tools/llm-cost/amazon/) : Provider hub with pricing, model lineup, and access details. - [Anthropic](https://aicost.tools/llm-cost/anthropic/) : Provider hub with pricing, model lineup, and access details. - [Baichuan](https://aicost.tools/llm-cost/baichuan/) : Provider hub with pricing, model lineup, and access details. - [Baidu](https://aicost.tools/llm-cost/baidu/) : Provider hub with pricing, model lineup, and access details. - [ByteDance (Doubao)](https://aicost.tools/llm-cost/bytedance/) : Provider hub with pricing, model lineup, and access details. - [Cohere](https://aicost.tools/llm-cost/cohere/) : Provider hub with pricing, model lineup, and access details. - [DeepSeek](https://aicost.tools/llm-cost/deepseek/) : Provider hub with pricing, model lineup, and access details. - [Google](https://aicost.tools/llm-cost/google/) : Provider hub with pricing, model lineup, and access details. - [Inception (Mercury)](https://aicost.tools/llm-cost/inception/) : Provider hub with pricing, model lineup, and access details. - [MiniMax](https://aicost.tools/llm-cost/minimax/) : Provider hub with pricing, model lineup, and access details. - [Mistral AI](https://aicost.tools/llm-cost/mistral/) : Provider hub with pricing, model lineup, and access details. - [Moonshot (Kimi)](https://aicost.tools/llm-cost/moonshot/) : Provider hub with pricing, model lineup, and access details. - [NVIDIA](https://aicost.tools/llm-cost/nvidia/) : Provider hub with pricing, model lineup, and access details. - [OpenAI](https://aicost.tools/llm-cost/openai/) : Provider hub with pricing, model lineup, and access details. - [Perplexity](https://aicost.tools/llm-cost/perplexity/) : Provider hub with pricing, model lineup, and access details. - [Reka](https://aicost.tools/llm-cost/reka/) : Provider hub with pricing, model lineup, and access details. - [Sakana AI](https://aicost.tools/llm-cost/sakana/) : Provider hub with pricing, model lineup, and access details. - [SpaceXSI](https://aicost.tools/llm-cost/xai/) : Provider hub with pricing, model lineup, and access details. - [Tencent (Hunyuan)](https://aicost.tools/llm-cost/tencent/) : Provider hub with pricing, model lineup, and access details. - [Upstage](https://aicost.tools/llm-cost/upstage/) : Provider hub with pricing, model lineup, and access details. - [Zhipu (Z.ai / GLM)](https://aicost.tools/llm-cost/zhipu/) : Provider hub with pricing, model lineup, and access details. ## Recently Verified Model Pricing - [Claude Fable 5.2](https://aicost.tools/llm-cost/anthropic/claude-fable-5-2/) : Anthropic; 1M context; pricing verified 2026-11-09. - [Claude Mythos 5.2](https://aicost.tools/llm-cost/anthropic/claude-mythos-5-2/) : Anthropic; 1M context; pricing verified 2026-11-09. - [Muse Ember](https://aicost.tools/llm-cost/meta/muse-ember/) : Meta; 1M context; pricing verified 2026-11-09. - [DeepSeek V4 Pro](https://aicost.tools/llm-cost/deepseek/deepseek-v4-pro/) : DeepSeek; 1M context; pricing verified 2026-11-08. - [GPT-5.6 Sol](https://aicost.tools/llm-cost/openai/gpt-5-6-sol/) : OpenAI; 1.05M context; pricing verified 2026-11-07. - [Hunyuan Hy4](https://aicost.tools/llm-cost/tencent/hy4/) : Tencent (Hunyuan); 1M context; pricing verified 2026-11-07. - [Qwen3.6 27B](https://aicost.tools/llm-cost/alibaba/qwen3-6-27b/) : Alibaba (Qwen); 256K context; pricing verified 2026-10-30. - [Qwen3.6 35B A3B](https://aicost.tools/llm-cost/alibaba/qwen3-6-35b-a3b/) : Alibaba (Qwen); 256K context; pricing verified 2026-10-30. - [Qwen3.6 Flash](https://aicost.tools/llm-cost/alibaba/qwen3-6-flash/) : Alibaba (Qwen); 1M context; pricing verified 2026-10-30. - [Qwen3.6 Plus](https://aicost.tools/llm-cost/alibaba/qwen3-6-plus/) : Alibaba (Qwen); 1M context; pricing verified 2026-10-30. - [Baichuan-M2](https://aicost.tools/llm-cost/baichuan/baichuan-m2/) : Baichuan; 32K context; pricing verified 2026-10-29. - [Baichuan-M2-Plus](https://aicost.tools/llm-cost/baichuan/baichuan-m2-plus/) : Baichuan; 32K context; pricing verified 2026-10-29. - [Baichuan-M3](https://aicost.tools/llm-cost/baichuan/baichuan-m3/) : Baichuan; 32K context; pricing verified 2026-10-29. - [Baichuan-M3-Plus](https://aicost.tools/llm-cost/baichuan/baichuan-m3-plus/) : Baichuan; 32K context; pricing verified 2026-10-29. - [Baichuan-Text-Embedding](https://aicost.tools/llm-cost/baichuan/baichuan-text-embedding/) : Baichuan; pricing verified 2026-10-29. - [Baichuan2-53B](https://aicost.tools/llm-cost/baichuan/baichuan-2-53b/) : Baichuan; 32K context; pricing verified 2026-10-29. - [Baichuan2-Turbo](https://aicost.tools/llm-cost/baichuan/baichuan-2-turbo/) : Baichuan; 32K context; pricing verified 2026-10-29. - [Baichuan3-Turbo](https://aicost.tools/llm-cost/baichuan/baichuan-3-turbo/) : Baichuan; 32K context; pricing verified 2026-10-29. - [Baichuan3-Turbo (128K)](https://aicost.tools/llm-cost/baichuan/baichuan-3-turbo-128k/) : Baichuan; 128K context; pricing verified 2026-10-29. - [Baichuan4](https://aicost.tools/llm-cost/baichuan/baichuan-4/) : Baichuan; 32K context; pricing verified 2026-10-29. - [Baichuan4 Air](https://aicost.tools/llm-cost/baichuan/baichuan-4-air/) : Baichuan; 32K context; pricing verified 2026-10-29. - [Baichuan4 Turbo](https://aicost.tools/llm-cost/baichuan/baichuan-4-turbo/) : Baichuan; 32K context; pricing verified 2026-10-29. - [Claude Fable 5](https://aicost.tools/llm-cost/anthropic/claude-fable-5/) : Anthropic; 1M context; pricing verified 2026-10-29. - [Claude Fable 5.1](https://aicost.tools/llm-cost/anthropic/claude-fable-5-1/) : Anthropic; 1M context; pricing verified 2026-10-29. ## Benchmark Leaderboards - [ARC-AGI-2 leaderboard](https://aicost.tools/benchmarks/arc-agi-2/) : GPT-6 Astra leads at 95% as of 2026-11-08; 105 systems ranked, each score linked to its source. - [ARC-AGI-1 leaderboard](https://aicost.tools/benchmarks/arc-agi-1/) : 6 systems tie for first at 98.5% as of 2026-11-08; 99 systems ranked, each score linked to its source. - [Terminal-Bench 4.0 leaderboard](https://aicost.tools/benchmarks/terminal-bench-4/) : Kimi K3.1 Pro leads at 65.7% as of 2026-11-08; 33 systems ranked, each score linked to its source. - [DeepSWE leaderboard](https://aicost.tools/benchmarks/deepswe/) : Gemini 4 Argon leads at 77.9% as of 2026-11-08; 27 systems ranked, each score linked to its source. - [ARC-AGI-3 leaderboard](https://aicost.tools/benchmarks/arc-agi-3/) : GPT-6 Astra leads at 62.7% as of 2026-11-08; 24 systems ranked, each score linked to its source. - [AutomationBench leaderboard](https://aicost.tools/benchmarks/automationbench/) : MiniMax M3.1 Flash leads at 62.8% as of 2026-11-08; 24 systems ranked, each score linked to its source. - [Agents' Last Exam leaderboard](https://aicost.tools/benchmarks/agents-last-exam/) : GPT-6 Astra leads at 59.3% as of 2026-11-08; 20 systems ranked, each score linked to its source. - [Humanity's Last Exam leaderboard](https://aicost.tools/benchmarks/hle/) : Claude Fable 5.2 leads at 74.2% as of 2026-11-08; 21 systems ranked, each score linked to its source. - [OSWorld 2 leaderboard](https://aicost.tools/benchmarks/osworld-2/) : Claude Opus 5.5 leads at 81.8% as of 2026-11-08; 20 systems ranked, each score linked to its source. - [Terminal-Bench 2 leaderboard](https://aicost.tools/benchmarks/terminal-bench-2/) : DeepSeek V4.1 Flash leads at 90.6% as of 2026-11-08; 18 systems ranked, each score linked to its source. - [FrontierCode leaderboard](https://aicost.tools/benchmarks/frontiercode/) : Claude Opus 5.5 leads at 54.4% as of 2026-11-08; 16 systems ranked, each score linked to its source. - [GPQA Diamond leaderboard](https://aicost.tools/benchmarks/gpqa-diamond/) : GPT-6 Astra leads at 96% as of 2026-11-08; 11 systems ranked, each score linked to its source. - [Chartography leaderboard](https://aicost.tools/benchmarks/chartography/) : Kimi K3.1 Pro leads at 90.2% as of 2026-11-08; 11 systems ranked, each score linked to its source. - [CharXiv Reasoning leaderboard](https://aicost.tools/benchmarks/charxiv-reasoning/) : Kimi K3.1 Pro leads at 89% as of 2026-11-08; 10 systems ranked, each score linked to its source. - [Terminal-Bench 3.0 leaderboard](https://aicost.tools/benchmarks/terminal-bench-3/) : DeepSeek V4.1 Flash leads at 30% as of 2026-11-08; 9 systems ranked, each score linked to its source. - [Terminal-Bench-Science leaderboard](https://aicost.tools/benchmarks/terminal-bench-science/) : Claude Mythos 5.2 leads at 75.3% as of 2026-11-08; 11 systems ranked, each score linked to its source. - [Toolathlon leaderboard](https://aicost.tools/benchmarks/toolathlon/) : MiMo-V2.6-Pro leads at 76.9% as of 2026-11-08; 9 systems ranked, each score linked to its source. - [CyberGym leaderboard](https://aicost.tools/benchmarks/cybergym/) : MiMo-V2.6-Flash leads at 95.1% as of 2026-11-08; 8 systems ranked, each score linked to its source. - [ExploitGym leaderboard](https://aicost.tools/benchmarks/exploitgym/) : GPT-6 Astra leads at 42.4% as of 2026-11-08; 8 systems ranked, each score linked to its source. - [NL2Repo-Bench leaderboard](https://aicost.tools/benchmarks/nl2repo/) : DeepSeek V4.1 Flash leads at 68.2% as of 2026-11-08; 8 systems ranked, each score linked to its source. - [OSWorld-Verified leaderboard](https://aicost.tools/benchmarks/osworld-verified/) : Kimi K3 leads at 84.8% as of 2026-11-08; 8 systems ranked, each score linked to its source. - [SWE-Bench Pro leaderboard](https://aicost.tools/benchmarks/swe-bench-pro/) : Gemini 3.6 Flash leads at 58.7% as of 2026-11-08; 8 systems ranked, each score linked to its source. - [CursorBench 4.0 leaderboard](https://aicost.tools/benchmarks/cursorbench-4/) : Claude Opus 5.5 leads at 57.8% as of 2026-11-08; 7 systems ranked, each score linked to its source. - [ExploitBench leaderboard](https://aicost.tools/benchmarks/exploitbench/) : GPT-6 Astra leads at 100% as of 2026-11-08; 7 systems ranked, each score linked to its source. - [GDP.pdf leaderboard](https://aicost.tools/benchmarks/gdp-pdf/) : Gemini 3.8 Flash leads at 35% as of 2026-11-08; 7 systems ranked, each score linked to its source. - [Harvey Legal Agent Benchmark leaderboard](https://aicost.tools/benchmarks/harvey-legal-agent/) : Kimi K3.1 Pro leads at 23.7% as of 2026-11-08; 7 systems ranked, each score linked to its source. - [MRCR v2 leaderboard](https://aicost.tools/benchmarks/gdm-mrcr-v2/) : Gemini 3.7 Flash leads at 97% as of 2026-11-08; 7 systems ranked, each score linked to its source. - [ProgramBench leaderboard](https://aicost.tools/benchmarks/programbench/) : Kimi K3.1 Pro leads at 31.5% as of 2026-11-08; 7 systems ranked, each score linked to its source. - [SEC-bench Pro leaderboard](https://aicost.tools/benchmarks/sec-bench-pro/) : GPT-6 Astra leads at 85.4% as of 2026-11-08; 7 systems ranked, each score linked to its source. - [APEX-Agents leaderboard](https://aicost.tools/benchmarks/apex-agents/) : Kimi K3.1 Pro leads at 69.4% as of 2026-11-08; 6 systems ranked, each score linked to its source. - [BrowseComp leaderboard](https://aicost.tools/benchmarks/browsecomp/) : Kimi K3.1 Pro leads at 92.1% as of 2026-11-08; 6 systems ranked, each score linked to its source. - [FrontierSWE leaderboard](https://aicost.tools/benchmarks/frontierswe/) : Claude Fable 5.2 leads at 69.9% as of 2026-11-08; 7 systems ranked, each score linked to its source. - [PostTrainBench leaderboard](https://aicost.tools/benchmarks/posttrainbench/) : Gemini 4 Argon leads at 45.3% as of 2026-11-08; 6 systems ranked, each score linked to its source. - [JobBench leaderboard](https://aicost.tools/benchmarks/jobbench/) : MiMo-V2.6-Pro leads at 62% as of 2026-11-08; 5 systems ranked, each score linked to its source. - [MLE-bench leaderboard](https://aicost.tools/benchmarks/mle-bench/) : Gemini 3.6 Flash leads at 63.9% as of 2026-11-08; 5 systems ranked, each score linked to its source. ## Glossary - [Agentic AI](https://aicost.tools/glossary/agentic-ai/) : Agentic AI is the class of AI systems in which language models work through several steps toward a goal — planning, calling tools and acting on results — rather than answering a single prompt once. - [AI agent](https://aicost.tools/glossary/ai-agent/) : An AI agent is a system in which a language model directs its own process — choosing tools, reading results and deciding the next step — until a task is done, instead of following a fixed script. - [AI hallucination](https://aicost.tools/glossary/ai-hallucination/) : An AI hallucination is output from a language model that is stated with confidence but is false, invented or unsupported by the sources the model was given. - [Batch API](https://aicost.tools/glossary/batch-api/) : A batch API is an asynchronous endpoint that takes many LLM requests as one job and returns the results within a set window, usually 24 hours, at a lower per-token rate than real-time calls. - [Blended rate](https://aicost.tools/glossary/blended-rate/) : A blended rate is a single per-million-token price that averages a model's input, output and cached-input rates under an assumed traffic mix, so models with different rate cards compare on one number. - [Cache write](https://aicost.tools/glossary/cache-write/) : A cache write is the request that first stores a prompt prefix in a provider's cache; some APIs bill those input tokens above the normal input rate, others bill them at the normal rate. - [Context window](https://aicost.tools/glossary/context-window/) : A context window is the maximum number of tokens a language model can work with in a single request — the prompt, the conversation so far and the answer it generates, counted together. - [Embeddings](https://aicost.tools/glossary/embeddings/) : Embeddings are lists of numbers — vectors — that an embedding model produces from text or other data, placed so that items with similar meaning end up close together and can be compared mathematically. - [Fine-tuning](https://aicost.tools/glossary/fine-tuning/) : Fine-tuning is further training of an already-trained model on a smaller set of task-specific examples, adjusting its weights so it follows a format, tone or narrow skill more reliably than prompting alone. - [Free tier](https://aicost.tools/glossary/free-tier/) : A free tier is a level of LLM API access that costs nothing per token, traded for low rate limits, fewer models and, at some vendors, looser terms on how your data is used. - [Inference](https://aicost.tools/glossary/inference/) : Inference is the act of running a trained AI model on new input to produce an output — answering a prompt, classifying an image — as opposed to training, which creates the model in the first place. - [Input vs output tokens](https://aicost.tools/glossary/input-vs-output-tokens/) : Input tokens are everything sent to a language model in a request, and output tokens are everything it generates in reply; APIs count and price the two directions separately, output usually higher. - [LLM](https://aicost.tools/glossary/llm/) : An LLM (large language model) is a neural network trained on very large amounts of text to predict the next token, which lets it generate and transform language — and, increasingly, code, images and tool calls. - [Long-context pricing](https://aicost.tools/glossary/long-context-pricing/) : Long-context pricing is a rate step some LLM vendors apply once a request's prompt passes a token threshold, billing that request at higher per-token rates than a shorter one. - [Max output tokens](https://aicost.tools/glossary/max-output-tokens/) : Max output tokens is the request setting that caps how many tokens a model may generate in one response — visible text plus any hidden reasoning — and the model-level ceiling that setting cannot exceed. - [MCP](https://aicost.tools/glossary/mcp/) : MCP (Model Context Protocol) is an open-source standard for connecting AI applications to external systems — tools, data sources and workflows — through a common interface any compatible client and server can share. - [Model deprecation](https://aicost.tools/glossary/model-deprecation/) : Model deprecation is a vendor's announcement that a model is no longer recommended and will be retired on a set date, after which requests to it fail; the vendor usually names a replacement. - [Model routing](https://aicost.tools/glossary/model-routing/) : Model routing is sending each request to one of several models — usually the cheapest one able to handle it — based on a rule or a classifier, instead of using one model for everything. - [Multimodal model](https://aicost.tools/glossary/multimodal-model/) : A multimodal model is an AI model that accepts more than one kind of input — text plus images, audio or video — converting each into tokens it can process and bill alongside the text. - [Off-peak pricing](https://aicost.tools/glossary/off-peak-pricing/) : Off-peak pricing is a time-of-day discount on an LLM API: the same model and request cost less per token when they run outside the vendor's published peak hours. - [Open-weight model](https://aicost.tools/glossary/open-weight-model/) : An open-weight model is an AI model whose trained parameters are published for anyone to download and run, under a licence that sets what you may do — without necessarily its training code or data. - [Price per million tokens](https://aicost.tools/glossary/price-per-million-tokens/) : Price per million tokens is the unit LLM APIs quote rates in: the dollars charged for one million input or output tokens, with input, output and cached input priced separately. - [Prompt caching](https://aicost.tools/glossary/prompt-caching/) : Prompt caching is a provider-side feature that stores the processed start of a prompt and reuses it on later requests with the same prefix, billing those input tokens at a discounted cached rate. - [Prompt engineering](https://aicost.tools/glossary/prompt-engineering/) : Prompt engineering is the practice of designing, testing and revising the instructions, examples and context you give a language model until its output reliably meets a defined standard. - [RAG](https://aicost.tools/glossary/rag/) : RAG (retrieval-augmented generation) is a technique that retrieves passages relevant to a question from a document collection and adds them to the model's prompt, so the answer draws on that material. - [Rate limits](https://aicost.tools/glossary/rate-limits/) : R
| – |
| 9 Nov 2026 |
| – | 9 Nov 2026 |