Last updated: 2026-09-30

Nesyona Research

Nesyona publishes original research on the AI tool economy. We track real pricing, free-tier longevity, capability boundaries, and cost-per-outcome across 200+ AI products. Our dated updates show the difference between marketed and delivered value.

Published research

2026-05-09 · published

AI API Token Price Decay 2022-2026: 12 Frontier Models Tracked

Per-model time-series of input and output token prices across 33 priced versions of 12 frontier model families (GPT-3.5 through GPT-5; Claude 1 through Claude Opus 4.5; Gemini 1.0 through 2.0 Flash; Llama 2/3/3.1 hosted; DeepSeek V2/V3; Mistral Large 2). Cost per a 1k-input/500-output reference task fell roughly 530x peak-to-floor between Mar 2023 and Aug 2024; Claude Opus held $15/$75 per million for 20 months across Opus 3 and Opus 4 before Opus 4.5 cut it 67%; open-weight hosted pricing leads each closed-weight cost cut by ~73 days median.

Dataset: 33 priced versions across 12 model families Primary sources: openai.com/api/pricing, anthropic.com/pricing, ai.google.dev/pricing, together.ai, fireworks.ai, api-docs.deepseek.com, mistral.ai, archive.org Wayback Machine, Stanford HAI AI Index, stateof.ai

2026-08-26 · published

The AI Recommendation Audit 2026: Which AI Tools AI Assistants Actually Recommend

We asked 4 live AI engines the same buying questions across 10 AI-tool categories (550 recommendations). They agreed on the #1 tool in 20% of categories. Per-tool AI-Visibility Scores, and a check of whether the tools they recommend still exist.

Dataset: 550 recommendations, 256 tools, 10 categories Engines: GPT-OSS, Qwen 3.6, Cohere Command-A, Gemini 2.5 Flash

2026-06-17 · published

The LLMOps Stack: OpenTelemetry, Self-Host and Pricing-Model Comparison 2026

A vendor-neutral comparison of 19 production LLM-ops tools across four stack layers on OpenTelemetry support, self-host, open-source licensing, pricing model, and built-in eval/guardrails. Open dataset, CC-BY 4.0, DOI-backed.

Dataset: 19 tools across 4 stack layers

2026-06-17 · published

Vector Databases for RAG: Deployment-Shape, License, Hybrid-Search and Pricing Comparison 2026

A vendor-neutral comparison of 11 vector databases for retrieval-augmented generation across five dimensions: deployment shape, open-source license, native hybrid search, metadata filtering, and pricing model. Open dataset, CC-BY 4.0, DOI-backed.

Dataset: 11 vector databases, 5 dimensions

statistics hub

AI Tools Statistics 2026

70 sourced statistics on the AI tools landscape: consumer and enterprise adoption, API token economics, frontier model release cadence, open vs closed-source share, deployment patterns, tool spend, labor productivity, and compute costs.

Format: each statistic linked to its source

Upcoming research

expected Q4 2026

AI Tool Pricing Reality 2026: Real Cost-per-Outcome Across AI Tools

In progress. Methodology being finalized.

How we work

Every study ships with full methodology disclosure: dataset size, time range, exclusions, and source attribution. Surveys disclose sample size, recruitment platform, and demographic stratification. We publish raw data alongside the analysis under a Creative Commons Attribution 4.0 license so researchers, journalists, and operators can verify our claims and build on the work.

Research is never sponsored or vendor-funded, and no one can pay to change a ranking or a finding. Every study is funded directly by Nesyona. Some pages carry one clearly labelled Featured slot, which is separate from the research and does not affect it.

Press and citations

Journalists, researchers, and operators are welcome to cite our findings under CC-BY 4.0 with attribution. For interviews, exclusive embargoed access, or additional data cuts, contact vincent@deepsynthesis.org.

Each study includes a press kit at its /press/ subpage with quotable findings, downloadable charts, and methodology summaries. Embargo requests honored when received at least 72 hours before scheduled publication.

All Nesyona research is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0). Attribution required: link to the study URL or to nesyona.com/research/.