Self-hosted
Free
Apache-2.0; infrastructure/provider fees separate
- Included scope
- Gateway; 4B compressor; CPU tool filter; recall
Paritok
● Identity checkedRecoverable agent context compression · Official-source review 2026-10-09
Paritok sits between coding agents and model providers, filtering tool schemas, compressing file/tool output and summarizing older history. Modified segments retain references so agents can retrieve original content when needed.
- Category
- Recoverable agent context compression
- Free access
- Apache-2.0 self-hosted gateway and 4B model are free software; hosted tool-schema filtering is free and new accounts receive one-time $5 credit.
- API access
- Gateway integrates through model base URLs and OpenAI-compatible upstreams, with read_original and tool-search mechanisms.
Is Paritok right for you?
A good fit for
Relevant to developers measuring context costs while retaining access to original source segments.
Before you choose
Hosted GPU compression costs $0.30 per million billable file-content tokens with o200k_base counting; prepaid top-ups start at $5. Provider inference and self-hosted hardware remain separate. Q4 model is about 2.5 GB; vendor suggests 8-GB GPU, while tool filtering can use CPU alone. Compression is lossy on the wire despite recoverable originals. Vendor reports 86.5% quality retention in a raw-model benchmark without recall; savings and quality were not reproduced, and flat provider subscriptions may not yield monetary savings. Hosted compression processes transmitted content; verify retention. No gateway was installed or benchmark run.
Workflow and handoff
Use a fixed fictional repository task, compare original and compressed outputs with recall enabled, and measure provider plus compression costs.
Pricing and practical limits
Hosted GPU compression costs $0.30 per million billable file-content tokens with o200k_base counting; prepaid top-ups start at $5. Provider inference and self-hosted hardware remain separate. Q4 model is about 2.5 GB; vendor suggests 8-GB GPU, while tool filtering can use CPU alone. Compression is lossy on the wire despite recoverable originals. Vendor reports 86.5% quality retention in a raw-model benchmark without recall; savings and quality were not reproduced, and flat provider subscriptions may not yield monetary savings. Hosted compression processes transmitted content; verify retention. No gateway was installed or benchmark run.
Review method
Official product evidence reviewed on 2026-10-09; product artwork inspected visually. No hands-on vendor application test was performed. Application performance, security claims and plan enforcement were not independently tested.
What it can do
Features & capabilities
Unknown is different from unavailable. Each fact carries its own evidence.
| Capability | Value | Evidence | Checked |
|---|---|---|---|
| Product workflow | Paritok sits between coding agents and model providers, filtering tool schemas, compressing file/tool output and summarizing older history. Modified segments retain references so agents can retrieve original content when needed. | Facts sourced | 2026-10-09 |
| Pricing and restrictions | Hosted GPU compression costs $0.30 per million billable file-content tokens with o200k_base counting; prepaid top-ups start at $5. Provider inference and self-hosted hardware remain separate. Q4 model is about 2.5 GB; vendor suggests 8-GB GPU, while tool filtering can use CPU alone. Compression is lossy on the wire despite recoverable originals. Vendor reports 86.5% quality retention in a raw-model benchmark without recall; savings and quality were not reproduced, and flat provider subscriptions may not yield monetary savings. Hosted compression processes transmitted content; verify retention. No gateway was installed or benchmark run. | Facts sourced | 2026-10-09 |
Understand the total cost
Paritok pricing & plans
Hosted GPU
Not listed
$0.30/million billable file tokens; prepaid from $5
- Included scope
- Managed file compression; free schema filtering; $5 initial credit
Alternatives to Paritok
View all ↗
Aura by NaridonAura provides an open-source desktop, CLI and MCP workspace for coding agents. It advertises semantic change history, plain-language diffs, function-level rewind and intent checks, with paid synchronization and collaboration features.
Explore tool
MermailMermail provides dedicated email inboxes for AI agents with business context, support automation and reviewed outbound drafts. An optional workspace wallet grants scoped access to supported purchases, with activity history and separately managed payment approvals.
Explore tool
RunInfraRunInfra provides hosted open models through OpenAI- and Anthropic-compatible APIs. It offers prepaid token billing and workspace coding plans, with session-aware caching and separately reported cached-token usage.
Explore toolThe practical questions
Frequently asked questions
Was the application hands-on tested?
No. This listing is based on dated official-source evidence and a visual artwork check. Unconfirmed costs and restrictions are identified explicitly.
What is Paritok used for?
Paritok sits between coding agents and model providers, filtering tool schemas, compressing file/tool output and summarizing older history. Modified segments retain references so agents can retrieve original content when needed.
Does Paritok have a free plan?
Apache-2.0 self-hosted gateway and 4B model are free software; hosted tool-schema filtering is free and new accounts receive one-time $5 credit.. This record lists ongoing free access; check the plan limits before starting.
How much does Paritok cost?
No paid monthly price is listed; this record treats the product as free to start. See the plan cards for entitlements, billing commitments and seat minimums.
Can I use Paritok through an API?
Gateway integrates through model base URLs and OpenAI-compatible upstreams, with read_original and tool-search mechanisms.. API access and subscription access may have different terms; consult the linked sources.
What should I check before choosing it?
Hosted GPU compression costs $0.30 per million billable file-content tokens with o200k_base counting; prepaid top-ups start at $5. Provider inference and self-hosted hardware remain separate. Q4 model is about 2.5 GB; vendor suggests 8-GB GPU, while tool filtering can use CPU alone. Compression is lossy on the wire despite recoverable originals. Vendor reports 86.5% quality retention in a raw-model benchmark without recall; savings and quality were not reproduced, and flat provider subscriptions may not yield monetary savings. Hosted compression processes transmitted content; verify retention. No gateway was installed or benchmark run.
Price history
No retained pricing changes yet. A current price alone does not establish a historical trend.
How this profile is supported
Official product information
Current official product source read individually; product artwork inspected visually.
Read original source ↗Pricing review
Official pricing and billing restrictions read individually.
Read original source ↗Facts apply to the named version and check date. Send a sourced correction if something changed.