Yaroslav Vikhariev
Yaroslav Vikhariev
Yaroslav Vikhariev is the founder of AI Cost & Tools Hub. He started this site because the pricing pages of the major LLM vendors aren't designed for the person paying the bill -- they're designed for the person selling you the model. Everything here is verified against the vendor's own canonical pricing page on the date stamped at the bottom, and built to be useful to whoever's actually doing the cost math.
Yaroslav started AI Cost & Tools Hub in 2026, after months of unable to find a single trustworthy pricing comparison of frontier LLMs. Every existing source was either vendor marketing or a one-off blog post running three prompts and calling it analysis. The gap was obvious; the project is the answer.
The site's editorial promise is narrow and verifiable: every price on every model page is checked against the vendor's own canonical pricing page on the date stamped at the bottom of that page. When a vendor changes a price, we re-verify and stamp a new date. When we don't yet have a data point, we leave the field blank rather than fabricate.
The project is independent and self-funded. No vendor pays for coverage, no model gets priority because of a commercial relationship, and the prices you see are the same prices anyone else paying the bill would see.
Coverage areas.
LLM API pricing
Per-model pages with input/output/cached rates, blended-cost scenarios, and price history — every number verified against the vendor's own pricing page on the date stamped at the bottom.
Provider profiles
Hub pages for OpenAI, Anthropic, Google, xAI, DeepSeek, Mistral plus seven frontier Chinese vendors — lineup, comparison table, release timeline, access points.
Methodology
Manual verification cadence. Every price has a source URL plus a date stamp. When data is missing or stale, the field is left blank — never fabricated. Full methodology →
Cost calculators
Per-use-case calculators — agent loops, RAG pipelines, long-document analysis, chat workloads. Phase 2 of the build plan.
Side-by-side comparisons
A-vs-B model and tool comparisons across pricing, features, benchmarks, and use-case fit. Phase 3 of the build plan.
Image gen, training research
No coverage of Midjourney, DALL-E, Runway, or other image/video generators. No coverage of ML training methods or architecture papers. Stays in lane: API economics and tool selection.
Recent filings.
What he can't accept
- Vendor-paid coverage. Sponsored content is not written by editorial staff.
- Free product samples conditional on positive reviews.
- Speaking fees from companies whose tools we rank.
- Equity in any AI tool company we cover.
- OK: press passes to AI Engineer Summit, NeurIPS, etc. (declared in coverage).
- OK: open-source contributions to tools we cover (declared on profile pages).
Conflicts of interest
Currently none. The site is independent and self-funded. No vendor relationships beyond the public pricing pages we cite, no equity in any AI tool company, no paid sponsorships.
If revenue infrastructure ships later — display ads or affiliate links per the published roadmap — it will be disclosed prominently at that time, and never as a condition of coverage. The editorial promise (vendor-canonical verification, blank fields over fabrication) does not change based on funding source.
Get in touch.
Story tips, corrections, sourcing questions. Direct contact preferred over DMs.