FAQ & Documentation

ZeroThink can use OpenZero as a local/self-hosted AI lane where the bridge is configured. Shared gateway access should stay limited; continued use belongs on a user-owned OpenZero node key saved in Neural Vault.

OpenZero nodes may support Z-Spark, a custom draft/verify route where a small local model drafts and the active model verifies or rewrites. It falls back safely when a node is not ready.

OpenZero can connect to Voicebox when a user installs that local voice service separately. Voice features remain optional and should be enabled only on machines prepared for voice workflows.

PDF requests work best when the user states audience, sections, evidence needs, caveats, and export purpose. Serious reports should be reviewed before sharing.

High-speed access to open-weights models. Ideal for coding and rapid inquiry.

• Groq Cloud: GPT-OSS 120B/20B, Compound, Qwen 3.6, Llama 3.1 8B, ALLaM 2
• NVIDIA NIM: Qwen 3 Coder 480B, Gemma 4 31B, DeepSeek V3.2, Codestral, Devstral
• CLI: Linked-account login or no-Google direct mode, folder-only or machine-wide access, SSH profiles, remote file workspace tools, Playwright MCP, named agents, autopilot runs, local plan board, verified self-update, local or Vault-backed provider keys, and local Z-Pepper password protection for files or folders
• Quantum Lane: Pro only
• ZeroThink Logic: Disabled

Full access to ZeroThink Logic, Pro routing, Quantum Zero, and proprietary closed-source model lanes.

• ZeroThink Logic: Active (Reasoning Layer)
• xAI: Grok 4.3, Grok Build 0.1, Grok Imagine API
• OpenAI: GPT-5.5, GPT-5.4, GPT-5.4 Mini, GPT-5.4 Nano, GPT-5 family fallbacks
• Google: Gemini 3.5 Flash, Gemini 3.1 Pro, Gemini 3.1 Flash-Lite, Gemini 2.5 fallbacks
• OpenZero: when the shared gateway is enabled, 5 shared local messages on ZeroThink, then bring your own registered OpenZero node key for continued local/self-hosted use
• Featherless: hosted large-model lane for TalkToAI ZERO, GLM 5.2, GLM 4.5, Qwen3 Coder, DeepSeek, and Kimi using the user's own Featherless key
• Quantum: Groq GPT-OSS 120B with IonQ Quantum Cloud telemetry and optional simulator job proof
• Billing: Pay with PayPal or verify a Solana wallet holding at least $10 worth of $AI. Token-holder access uses a signed wallet message, never asks for a seed phrase, and is automatically rechecked. If holdings are sold below the threshold, token-holder Pro access is removed. Monthly payments are non-refundable. Yearly refund requests are within 7 days via shaf@talktoai.org

Hold at least $10 worth of $AI in a verified Solana wallet to activate ZeroThink Pro access. This unlocks the same Pro surface: ZeroThink Logic, Pro model lanes, Quantum Zero, Oracle improvements, and the wider builder stack.

Verification uses a one-time wallet signature. ZeroThink never asks for seed phrases, never moves tokens, and never requires a payment transaction for token-holder access. The wallet only signs a message proving ownership.

Token-holder Pro is monitored. The server rechecks live holdings and live market value. If the wallet sells below the threshold, token-holder Pro access is automatically removed on the next scheduled check. PayPal Pro accounts are not affected by this token check.

Use case: the coin is an access key for an AI product, not just a meme. Holding enough $AI unlocks a real software benefit, and selling below the floor removes that benefit. That creates a direct utility loop between the token and ZeroThink Pro.

$AI Token CA: dHdkhZsmBQhXjQoKbkyajZ3Ff4W2XumGyVJX4jypump

Verify holder access on the Upgrade page.

Track AI/SOL on GeckoTerminal.

$AI holder access is a product utility feature. It is not financial advice, not a promise of profit, and not a guarantee of token price. Access depends on live wallet balance, market data, and the current ZeroThink eligibility rules.

The Oracle route is tuned for both crypto scenarios and broader strategic calls. Modes include crypto, general, launch, and risk.

Oracle now defaults to Groq + GPT-OSS 120B for a strong speed / reasoning balance, while still letting you switch provider and model directly from the desk. Current Groq alternatives include Compound, GPT-OSS 20B, Qwen 3.6, Llama 3.1 8B, and ALLaM 2; older Groq IDs are kept only as migration aliases.

The Oracle page includes a Solana support card so people who use the desk heavily can donate directly to the ZeroThink build.

If someone wants ZeroThink without Google login first, the lean browser route is exposed at /guest. That lane accepts local Groq or NVIDIA keys, keeps memory/history off, does not connect to the ZeroThink account/Vault, and stores the key only in the browser.

Quantum is a special Pro agent inside Studio. It does not replace the normal default route. Instead, it switches into a separate research lane that reasons through Groq GPT-OSS 120B and then grounds that reasoning with IonQ Quantum Cloud account data, backend telemetry, and small simulator job records when enabled, so the answer reflects real quantum infrastructure instead of vague sci-fi language.

ZeroThink Pro is not claiming that the LLM itself runs on quantum hardware. It is hybrid quantum-informed AI: a classical model does the language reasoning, while real quantum-cloud telemetry, backend/account context, and optional IonQ simulator job records are brought into the prompt/context layer. Real integration, not fake magic.

The lane is designed as a research surface, not a fake “magic quantum oracle.” ZeroThink uses IonQ key validation, backend availability, queue posture, characterization context, and job IDs/results when a simulator proof is created, while Groq remains the fast reasoning core. In short: classical AI engine, quantum-informed research mode.

Quantum Zero and Studio Quantum are tuned to stay friendly to free or low-volume IonQ accounts. Packet building is local/offline, backend telemetry is cached for 30 minutes, job checks are cached for 60 seconds, Studio chat reuses recent simulator proof jobs by default, and hardware QPU submissions require a deliberate checkbox.

IonQ Quantum needs two keys: a Groq key for GPT-OSS 120B and an IonQ key for the quantum cloud lane. IBM Qiskit needs your Groq key for the answer and an optional IBM Quantum token when you want IBM Runtime/backend context. Save keys in the Neural Vault, then switch the provider pill to QUANTUM or QISKIT.

Quantum access is reserved for Pro. Free accounts stay on the fast open-weights lanes and do not unlock the IonQ-backed agent.

The dedicated Quantum Zero page is the simple chat surface. It now includes a larger real experiment deck for Pro users: Bell Pair Correlation, GHZ Shared-State Pulse, Entropy Beacon, Parity Pattern, Delta Branch Ethics, Omega Phi Xi Balance, ADEV Decision Spread, Harmonic Base Matrix, Six Layer Decision Pattern, and prompt-seeded QuantumZero packets. These submit/reuse small IonQ simulator jobs, read the probability histogram, then explain dominant state, top states, entropy/spread, backend evidence, and a practical result interpretation inside ZeroThink so users do not need to leave the platform. The advanced Quantum Lab restores the deeper operator console: packet inspection, IonQ status checks, proof jobs, job-result review, telemetry forensics, and research-safe mission templates. IonQ simulator submission creates a real quantum-cloud job record when requested; it does not claim magical broadcasting or qubits running the chatbot.

Quantum Zero includes a provider-backed simulator research chain seeded from private Zero themes without exposing private equations, raw signals, run identifiers, or internal ratios. The public lane describes what the chain is for: controlled quantum-cloud experiments, entropy checks, balance checks, decision-spread checks, harmonic packets, and reproducible research notes. It is evidence for the research workflow, not proof of remote activation.

The Qiskit lane is separate from IonQ. It builds a calculation packet first, then lets Groq reason over that packet. Supported starter algorithms are Bell correlation, GHZ shared state, Grover search, QAOA MaxCut, a small VQE-style ansatz, QFT phase spread, and prompt-seeded Zero pattern circuits. Local Qiskit simulation can run without an IBM token; IBM Runtime/backend context requires the user's own IBM Quantum token and optional instance/CRN. Hardware job submission should remain deliberate and clearly labelled.

The ZeroThink CLI, aka Agent Zero, installs on Windows PowerShell, macOS, and Linux. It can pair to the same ZeroThink account as the web app or run in a no-Google direct mode with a local Groq or NVIDIA key. Its operator surface includes guided setup, scoped file access, local and remote workspaces, SSH/SFTP, checkpoints, planning, browser research, Playwright/MCP, encrypted file bundles, PDF report creation, saved-server health checks, site starters, playable browser game starters, named agents, and evidence-first research creation. New research commands can inspect approved files on the local computer or a saved server profile, then create a paper, article, survey, review, or proposal with a source ledger, claim matrix, verification flags, and Markdown export. The terminal remains concise; the richer public ZeroThink Reasoning Brief is shown in Studio.

The CLI supports direct file injection, workspace scans, file search, file writes and replacements, zip creation and extraction, Z-Pepper file/folder encryption and decryption, automatic encrypted project bundles, lock/protect/unlock aliases, remote SCP transfers, remote file inspection and edits on saved server profiles, repo-level guidance files like ZEROTHINK.md, CLAUDE.md, AGENTS.md, and .claude/CLAUDE.md, local web search, saved browser pages, Playwright MCP automation for dynamic pages, page attach flows, and the same packed file strategy used in the web app, so large source files and config dumps are easier to analyze without immediately blowing the window. The CLI reasoning layer now follows a stronger private orient → inspect → plan → act → verify → report workflow instead of staying in generic assistant mode.

The public install and manual page is exposed at /cli. The CLI can run in a safer folder-only mode for one repo or switch to machine mode when you want shell and server operator access, and the guide now walks beginners through install, linked-account mode, no-Google direct mode, first prompt, help topics, model selection, autocomplete, local checkpoint/resume flows, local planning, and the limits of direct mode. It also has a built-in zerothink update command that verifies the published CLI hash before replacing the local script, zerothink status for a quick local readiness snapshot, a remote workspace lane via zerothink remote ..., saved-profile checks like zerothink server check prod /var/www/html, local PDF reports like zerothink pdf create report.pdf --file notes.md, a local MCP layer with a built-in Playwright browser profile via zerothink mcp add-playwright and zerothink playwright open https://example.com, plus self-driven agent workflows like zerothink autopilot "audit this repo", zerothink agent create zero1 "map the auth flow", zerothink agent log zero1, zerothink stop zero1, and in-prompt /stop. For coding, the current recommended defaults are openai/gpt-oss-120b on Groq and qwen/qwen3-coder-480b-a35b-instruct on NVIDIA, and the fastest way to inspect the curated shortlist is zerothink models coding. If you just type a normal task like make a small game or make a new project folder for shafaet.com inside the interactive prompt, Agent Zero now routes that into execution mode automatically instead of staying in pure chat mode, unless you switch to a staging lane like /steer plan. The file/tree tools now keep broad inspections compact by skipping common dependency, cache, vendor, and backup directories unless you explicitly inspect those paths.

If you skip Google login in the CLI, ZeroThink can still auto-detect GROQ_API_KEY or NVIDIA_API_KEY from your local shell, or you can store them locally with zerothink key set groq / zerothink key set nvidia. That lane does not import Vault keys, memory, saved history, linked-account session state, or Pro account state. Local CLI browser tools like zerothink search, zerothink browse, /search-web, /page, and local Playwright MCP browser commands still work because they are machine-local terminal helpers rather than website account features. For Groq and NVIDIA, the CLI can also fall back to that local direct-key lane automatically if the linked provider lane fails. If you ever want the exact live route instead of asking the model to guess, use zerothink runtime or /runtime. If you already have the CLI installed, the safe update command is simply zerothink update.

Research Paper Creator provides separate stages for search protocol, evidence plan, source ledger, outline, draft, reviewer critique, and revision. It supports paper type, method, audience, citation style, date range, source policy, inclusion/exclusion criteria, uploaded research files, known sources, integrity checks, and project export.

Optional uploads such as TXT, Markdown, CSV, JSON, HTML, RTF, DOCX, legacy DOC, BibTeX, notes, abstracts, draft articles, and supervisor comments are read in the browser and attached only to the current generation request. They are not saved as a public research library upload.

In Studio, prompts like create a PDF report about this trigger ZeroThink's AI PDF artifact mode: the selected model drafts a polished document, optional WEB mode or explicit research wording can add Serper search context, and the answer returns a private download button. PDF Creator remains the manual editor/exporter for pasted notes, reports, drafts, source summaries, or Paper Creator output. Uploaded text files are read in the browser and sent only as document text when the user creates the PDF; the endpoint does not accept server file paths or browser-supplied URLs.

The CLI can inspect only the files or folders the user explicitly supplies. Text, code, CSV, Markdown, DOCX, and optional PDF extraction are supported. Common credential files, private keys, dependency folders, and oversized files are excluded or inventoried by default.

In machine mode, a saved SSH profile can supply remote research files without publishing the server address in documentation. The CLI reads through SFTP, applies the same exclusions, creates a visible inventory, and removes its temporary evidence pack after the model call.

Choose protocol, evidence, outline, draft, critique, revision, or full. Use --dry-run to review scope without a model call, --library for compact private Zero Library retrieval, and --web only when live discovery is needed.

ZeroThink must not invent author names, papers, quotations, DOIs, URLs, page numbers, statistics, or results. Search snippets are discovery leads, not proof of full-text access. Unverified claims remain visibly marked, and the final export includes the supplied source inventory.

The source ledger separates Uploaded Source, User Note, Retrieved Search Result, Full Text Supplied, Model Candidate Lead, and Needs Verification. Candidate leads are not citations. They are search targets until a DOI, URL, database result, or supplied file verifies them.

Yes. zerothink freewebpanel, with the shorter alias zerothink fwp, can install, manage, inspect, back up, diagnose, repair, update, and help operate a FreeWebPanel server. Start with status, check, or health, then use backup-first plans for repair, update, and deploy. It includes web-server, PHP, database, mail, DNS, SSL, permission, firewall, redacted-log, and existing-panel takeover/migration checks without publishing private infrastructure examples.

No. Installation is one feature. ZeroThink CLI is the wider local server-management and hosting-operations layer for FreeWebPanel and related TalkToAI/ZeroThink work. It supports health, config, domain, log, permission, Apache/PHP, SSL, backup, repair, update, and deployment workflows after installation.

Inspection commands are read-only. Install, repair, update, deploy, and backup commands show their plan first. Use --dry-run for an explicit preview. A change requires --yes, and repair, update, or deploy should normally include --backup-first.

Common password, secret, token, private-key, database, SMTP, and API-key values are redacted from configuration and log previews by default. Use --no-redact only in a trusted local terminal when the raw value is genuinely required.

Ubuntu is the primary install lane. AlmaLinux, Rocky Linux, and RHEL-family servers are supported through the RHEL-family lane. OpenZero and ZeroMint AIOS servers use the OpenZero AIOS lane. Existing-panel takeover is treated as a migration workflow: run preflight first, verify off-server backups, then approve the replacement plan deliberately.

Start with zerothink fwp status, then zerothink fwp doctor --verbose. For automation use zerothink fwp health --json. Before maintenance, preview zerothink fwp backup --path /opt/freewebpanel --dry-run.

Upload notes, code, logs, docs, FASTA/FASTQ files, raw genotype exports, VCF files, or project references at /library. Each library is account-private: files live in a user-scoped server directory, the public web cannot browse storage, and search only reads that user's own index. In Studio or the CLI, mention @library when you want Agent Zero to attach only the most relevant snippets.

/dna is now a one-click DNA evidence lab. Users upload a MyHeritage, 23andMe, Ancestry-style, CSV/TSV, raw, genome, or VCF file, then press Run Full DNA Report. ZeroThink returns plain-English cards, a signal wheel, an Evidence Ledger, Markdown/PDF export, and a DNA Agent follow-up box that receives summaries only, not raw genotype rows. Under the hood, ZDNA-MAP v1.1 scores raw-file quality, guesses genome build conservatively, tracks duplicates/chromosome coverage, prepares health research evidence cards, creates confidence-bounded heritage/genetic-affinity lenses, and can build optional GEDCOM/story-based Lineage Clues. The local data-pack path supports ClinVar, GWAS Catalog, PharmGKB/CPIC, HPO, UCSC liftover, 1000 Genomes/gnomAD-lite reduced panels, and Wikidata lineage subsets when installed. The Health lane does not diagnose or advise treatment. The Heritage lane does not invent exact ethnicity percentages or identity claims. The Lineage lane does not prove royal ancestry from DNA alone.

/media is now a downloadable Media Studio: AI voice MP3 generation, GPT Image PNG generation, xAI Grok Imagine Pro video jobs, SeedDance / BytePlus Ark cinematic video tasks, legacy Sora fallback, MP4 download links, and a browser-local asset shelf. Video generation is Pro-only; users save xAI and SeedDance keys in Vault Keys when they want their own provider billing.

Agent Zero includes the same idea locally: zerothink library add ., zerothink library search "query", zerothink library context "bug", zerothink research paper "topic" --source ./papers --library, and zerothink dna analyze sample.fasta. Deeper AlphaGenome execution stays on the website/server side because biological API keys should remain centrally controlled rather than scattered across user terminals.

The library accepts only safe text/code/sequence file types, blocks direct storage browsing, ignores dangerous archive execution paths, and keeps generated context compact. It helps the agent read better; it does not execute uploaded files.

The formal core is Z/9Z, the triadic subgroup {0,3,6}, rotation x -> x + 3 mod 9, and the mirror operator x -> 8x mod 9 = -x mod 9. This keeps the agent honest and useful.

ZMath now answers like a normal agent first, then places the lab tools underneath. It returns parsed residues, balanced notation, reset/mirror traces, deterministic ZM9 fingerprints, state-machine JSON, practical software patterns, file profiles, and downloadable markdown reports. It is useful for workflow design, reset logic, checksum tags, reversible UI states, encryption handoff metadata, and symbolic research organization.

The ZMath page can create portable .zme1 files locally in the browser. It uses browser WebCrypto with AES-GCM-256 and PBKDF2-SHA256, plus optional pattern-image and Z-Pepper factors. The password, pattern file, and plaintext are not uploaded to ZeroThink. The ZMath packet stored in metadata is for deterministic workflow/fingerprint context, not a claim of extra cryptographic magic.

The -0, +0, and infinity labels are state annotations layered on top of modular arithmetic. ZMath does not pretend they are standard numeric residues; it uses them to preserve transition direction in protocols and research notation.

States the real problem, constraints, success condition, strongest available facts, retrieved context, calculations, and source limitations.

Surfaces the strongest credible alternative or failure mode, then identifies what remains unknown, unverified, assumed, or time-sensitive.

Checks likely harms, stakeholders, reversibility, and downstream effects before stating the chosen approach and one concrete next step.