Touchmark is live · pricing from $4,165.33/1 b-token · last verified 22 Oct 2026 · viability monitor
Touchmark
Touchmark is a live order book for buying and selling reserved AI inference capacity at forward-curve prices.
A real forward market for tokens, which is genuinely different from signing a private reserved-capacity deal. The formula is public, delivery months sit side by side, and buyers pay no transaction fees. The trade is firm: sales are final, unused Tokens expire at window close, and you're committing before you know your actual volume.
Verdict generatedVerified 8d ago · liveness 52/100 · cite: rightaichoice.com/tools/touchmark
- Engineering and procurement leads at inference-heavy companies who can forecast token volume weeks ahead
- AI startups that want reserved-capacity pricing without a private deal they can't exit
- Teams willing to trade a firm forward commitment for a posted price and up to 20% off list
- GPU providers with idle hours who want to set a list rate, a floor and commit revenue at sale
- Teams whose usage swings month to month with no reliable volume forecast
- Buyers who can't accept final sales and Tokens expiring at delivery window close
- Anyone wanting a no-commitment per-token rate card instead of a forward block
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Touchmark if your monthly token burn swings too unpredictably to forecast weeks ahead, or if you need a negotiated MSA and custom legal terms — sales are final and unused Tokens expire at delivery window close.
Input and cache tokens aren't free draws — they count against your block at the listing ratio, so GLM 5.2 input at ×0.32 means 1M input tokens consume 320,000 of your Tokens.
Touchmark's listed shelf is built for teams committing 1B-token blocks — a roughly $4K entry (GLM 5.2, Sep 1 – Oct 1 at $4,165.33) up to $18,000 for Kimi K3 Fast in the Nov 1 – Dec 1 window. That is meaningfully cheaper than the list rates those same models carry ($4,400 and $22,500), but it's a firm commitment rather than a rate card. Smaller buyers and anyone who can't forecast volume should stay on their lab's committed-use plan.
In short
Touchmark — Touchmark is a live order book for buying and selling reserved AI inference capacity at forward-curve prices. Best for Engineering and procurement leads at inference-heavy companies who can forecast token volume weeks ahead, AI startups that want reserved-capacity pricing without a private deal they can't exit, Teams willing to trade a firm forward commitment for a posted price and up to 20% off list. Plans from $4,165.33/1 b-token.
What is Touchmark?
Touchmark — Touchmark is a live order book for buying and selling reserved AI inference capacity at forward-curve prices. Best for Engineering and procurement leads at inference-heavy companies who can forecast token volume weeks ahead, AI startups that want reserved-capacity pricing without a private deal they can't exit, Teams willing to trade a firm forward commitment for a posted price and up to 20% off list. Plans from $4,165.33/1 b-token.
Touchmark runs a two-sided market for reserved AI inference capacity — the thing you'd normally negotiate as a private committed-use deal with a model lab. Buyers take fixed 1B-Token blocks on a named model for a specific delivery month at a posted ask, wire the block to one OpenAI-compatible API key, and draw it down at pinned per-token rates from the month's first day. The listed shelf has included GLM 5.2, GLM 5.3 Flash, GLM 5.2 Fast volume blocks from provider Wafer, Kimi K3 Fast, and Gemma 4 31B from OpenRelay. Unit accounting is explicit rather than hand-waved. One Token equals one output token; input counts at the model's input-to-output price ratio and cache reads at the cache-to-output ratio, so a ×0.32 input ratio means 1M input tokens draw 320,000 Tokens. Every listing publishes its own ratios, which is what lets you translate a block price into an effective cost per million raw tokens for your traffic mix. Pricing runs off a single published formula: list rate S × (1 − min(20%, 10% × τ/30)), where τ is days to the delivery month's maturity. The 20% cap binds from roughly 60 days out. A GLM 5.2 Sep 1 – Oct 1 block is quoted at $4,165.33 against a $4,400 list. Beyond the shelf, you can post a request for quote with your own terms — larger blocks, dedicated GPUs, reserved throughput, latency floors — and watch provider bids land, or reserve hourly throughput up to 3 hours ahead of demand. Who it's for: engineering and procurement leads at inference-heavy companies who can forecast volume weeks ahead, and GPU providers with idle hours to place. Touchmark reports 4T+ tokens of RFQ volume and is backed by Y Combinator. Compared with a lab's committed-use plan, the difference is transparency — the ask is public, the discount is computable, and trades reportedly clear 10–60% below retail with no buyer transaction fees.
Key features
- Live two-sided order book for reserved AI inference capacity with visible bids and asks
- Forward curve pricing: S × (1 − min(20%, 10% × τ/30)), cap binding ~60 days before month-end
- Buy fixed 1B-token blocks on named models for a chosen delivery month
- Hourly throughput listings — reserve peak capacity up to 3 hours ahead of demand
- Connect a block to one OpenAI-compatible API key and draw down at pinned per-token rates
- 1 Token = 1 output token; input counts at the input:output ratio, cache reads at cache:output
- Per-listing ratio disclosure (GLM 5.2 input ×0.32 / cache ×0.06; Kimi K3 Fast input ×0.20 / cache ×0.02)
- SLA-backed service levels: availability ≥ 99.0%, TTFT ≤ 1,000 ms p95, speed ≥ 60 t/s p05
- Resale of untouched contracts on consenting series before the first consumed token
- Benchmark-referenced Synthetic Token contracts that convert into the leading coding model's Tokens
- Top-1 model named by a public rule about 20 business days before the window opens
- Pre-release models quoted in synthetic units that settle into Tokens at launch
- Request for quote: post volume, dedicated GPUs, reserved throughput or latency floors and see every bid
- Provider-pool failover with SLA-backed buyer protections
- Provider-side listing: set your own list rate S and floor, deliver on your own endpoints
Behind the verdict
Touchmark only makes sense if your token burn is boring. Not small — boring. If you can look at the last three months and sketch next quarter within a band, the forward curve is a cheaper way to buy that baseline than paying retail on demand. If your usage is spiky or you're still finding product-market fit, this is a way to convert cash into expiring inventory. The thing we like most isn't the discount, it's the disclosure. Provider, list rate, delivery window, and the input/cache ratios all sit on the listing. That last part matters more than buyers realize: a ×0.20 input ratio and a ×0.32 input ratio are different products for a RAG-heavy workload, and you can actually compare them here instead of guessing from a lab's blended price. Where it bites is the terminal condition. Unused Tokens expire when the window ends. There's no rollover, no refund, no partial credit. Resale exists for untouched contracts on consenting series before the first consumed token, which is a real exit but not one you should plan around. Under-buy and you're back at retail for the overage; over-buy and you eat it. The hourly throughput channel is the underrated half. Reserving peak capacity up to 3 hours ahead with the same SLA protections (≥ 99.0% availability, TTFT ≤ 1,000 ms p95, ≥ 60 t/s p05) covers the demand curve you can't forecast a month out, and it's an easier first purchase than a full block if you want to test the pipes. Closest alternative is whatever committed-use plan your model lab offers directly, or a GPU cloud's reserved instance. Those give you a vendor relationship and a negotiated MSA; Touchmark gives you a price you can see and a market you can sell into. If procurement needs custom legal terms, this isn't the venue. If you want to know exactly what the discount
Models under the hood
as of 2026-10-24
Limitations
- Everything on Touchmark is a forward commitment: unused Tokens expire when the delivery window ends, and pricing follows a published curve that deepens as the delivery month recedes, capped at 20% below list (binding about 60 days before month end).
- Resale listings are seller-priced, and per the profile resale is only available on consenting series before the first consumed token.
- On Synthetic Token contracts the model is named later by a public selection rule, so the actual allocation isn't known at purchase; effective cost per 1M raw tokens also depends on your own traffic mix via per-listing input and cache ratios.
as of 2026-10-12
Researching Touchmark? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Is Touchmark still active?
Touchmark is live · pricing from $4,165.33/1 b-token · last verified 22 Oct 2026 · viability monitor.
Viability score
How well maintained and how widely used is Touchmark? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →What's new in Touchmark
Checked 9 days agoAcross the latest 1 update: 1 launch.
What do users say about Touchmark?
What people actually say about Touchmark — is it worth it?
Our community sentiment research written Jul 3, 2026 is more than 90 days old, so it is being re-reviewed against current data and is not shown.
How we verified this
Every listed fact about Touchmark— pricing, features, links, status — is re-read from the vendor's own pages by our verification pipeline and each change is logged with a date; the last pass ran on 22 Oct 2026. Pages mix model-assisted synthesis with human review; see how we verify.
Verification history
We have re-verified Touchmark 11 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — changed models, limitations, pricing page
- — changed our verdict, pricing tiers, description
- — changed who it suits, summary, features, pricing tiers, description, who should skip it, our verdict, limitations, models, pricing page
- — changed models, who should skip it, summary, our verdict, features, hidden costs, use cases, setup time, pricing fit, limitations, switching in, switching out, description
- — changed features, our verdict, summary, description, who it suits, pricing tiers, who should skip it
- — changed link health
Showing the 6 most recent of 11 verification passes.
Free to cite with attribution — this page re-verifies continuously.
How much does Touchmark cost in 2026?
Touchmark: Plans from $4,165.33/1 b-token. Pricing last verified 22 Oct 2026.
Plans compared
Every published Touchmark tier, its price as the vendor publishes it, who it fits, and what it adds over the tier below. The 12-month cost projection follows.
| Plan | Price | Billing | Who it's for |
|---|---|---|---|
| Volume contract — GLM 5.2 (Sep 1 – Oct 1 delivery) | $4,165.33/1 b-tokenVendor: “$4,165.33 per 1B-Token block” | usage | — |
| Volume contract — Kimi K3 Fast (Nov 1 – Dec 1 delivery) | $18,000/1 b-tokenVendor: “$18,000 per 1B-Token block” | usage | — |
| Top-1 floating series (benchmark-referenced) | $22,500/1B synthetic-unitVendor: “$22,500 per 1B synthetic-unit block (Oct 1 – Nov 1 window)” | usage | Buyers who want front-row coding-model capacity but don't want to bet on which model wins the published index.Adds: Benchmark-referenced rather than model-fixed: quoted in Synthetic units at $22,500 per 1B block for Oct 1 – Nov 1, resellable before the first consumed token. |
| Pre-release synthetic series — Kimi-X (Dec 1 – Jan 1 window) | $0.29 per synthetic unit | usage | — |
| Hourly throughput | Price not publishedVendor: “Curve-priced per listing (e.g. GLM 5.3 Flash at 1 MTok/min)” | — | — |
| Request for quote | Custom pricingVendor: “Custom — providers bid against your terms” | quote | Buyers whose needs the standard shelf doesn't list — larger blocks, dedicated GPUs, reserved throughput, or a latency floor.Adds: Custom terms with visible provider quotes: you see every quote as it lands, providers see only the level to beat, and accepting the best creates a contract. |
| Provider listing | Provider-set list rate S and floor | — | GPU providers with idle hours to place before a month starts who want revenue committed at sale rather than at usage.Adds: Supply-side tier: set your own list rate S and floor, keep delivery on your own endpoints, and see demand by delivery month before the month starts. |
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Where the pricing makes sense
The company stage and team size where Touchmark's pricing actually pencils out — and where peers do it cheaper.
Touchmark's listed shelf is built for teams committing 1B-token blocks — a roughly $4K entry (GLM 5.2, Sep 1 – Oct 1 at $4,165.33) up to $18,000 for Kimi K3 Fast in the Nov 1 – Dec 1 window. That is meaningfully cheaper than the list rates those same models carry ($4,400 and $22,500), but it's a firm commitment rather than a rate card. Smaller buyers and anyone who can't forecast volume should stay on their lab's committed-use plan.
Has Touchmark's pricing changed?
No plan price for Touchmark has moved since we began logging in Aug 2026; the pricing page is re-read on every verification pass.
Who should pick Touchmark?
Pick Touchmark if you are Engineering and procurement leads at inference-heavy companies who can forecast token volume weeks ahead, AI startups that want reserved-capacity pricing without a private deal they can't exit, Teams willing to trade a firm forward commitment for a posted price and up to 20% off list, GPU providers with idle hours who want to set a list rate, a floor and commit revenue at sale.
Look elsewhere if you are Teams whose usage swings month to month with no reliable volume forecast, Buyers who can't accept final sales and Tokens expiring at delivery window close, Anyone wanting a no-commitment per-token rate card instead of a forward block, Procurement teams that need a negotiated MSA and custom legal terms rather than an on-book contract.
Skip it if Skip Touchmark if your monthly token burn swings too unpredictably to forecast weeks ahead, or if you need a negotiated MSA and custom legal terms — sales are final and unused Tokens expire at delivery window close.
GeneratedReal-world workflow fit
Concrete scenarios for the personas Touchmark actually fits — and what changes day-one when you adopt it.
You have a Q3 launch and a solid token forecast. You open the book, compare the best ask across the next three delivery months, and buy a 1B-token GLM 5.2 block for the Sep 1 – Oct 1 window at $4,165.33 against a $4,400 list.
Outcome: You wire the block to one OpenAI-compatible API key and draw it down at pinned rates from September 1, with input counting at ×0.32 and cache reads at ×0.06.
The listed shelf doesn't carry the terms you need, so you post a request for quote with your volume, throughput, and latency requirements.
Outcome: Provider quotes land against a level rather than a name; accepting the best quote creates a contract on the spot. Touchmark reports 4+ trillion tokens requested through RFQ.
You have capacity coming free and no committed buyer. You list it forward, setting your own list rate S and floor, with delivery staying on your own endpoints.
Outcome: You see demand by delivery month before the month begins and commit revenue at sale rather than at usage, with Touchmark handling settlement.
Use cases
- Lock a fixed 1B-token block on a named model for a delivery month at a posted ask below list
- Compare the best ask across several delivery months before committing a quarter's inference budget
- Reserve peak capacity by the hour ahead of a scheduled spike, with SLA-backed protections
- Hedge a bet on the top coding model without knowing its name, using the benchmark-referenced Synthetic series
- Reserve capacity on a pre-release model quoted in synthetic units before it ships
- Post a request for quote when you need a larger block, dedicated GPUs, reserved throughput, or a latency floor
- List idle GPU hours forward as a provider to commit revenue at sale rather than at usage
- Read demand by delivery month before committing to a GPU buildout
Setup time & first value
How long it actually takes to get something useful out of Touchmark — broken out by persona, not the marketing-page minute.
For a standard volume block: minutes to compare asks and buy, then one API-key wiring step before the delivery window opens — first value lands on the window's first day. For an RFQ requiring dedicated GPUs, reserved throughput, or a latency floor, expect a longer cycle while provider quotes land and you accept. Provider-side listing is a self-serve rate-and-floor setup with delivery on your own
Switching to or from Touchmark
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From a lab committed-use plan: move a forecastable baseline of demand onto a posted 1B-token block and keep the lab plan for anything you can't forecast.
- →From per-token rate cards: model your traffic mix against the listing ratios (GLM 5.2 input ×0.32 / cache ×0.06) using the live calculator before committing a block.
- →From private reserved-capacity deals: replace an unseen single quote with a posted ask you can compare across delivery months.
- ↗To your lab's committed-use plan: accept a single negotiated rate and no resale, in exchange for dropping the forward-commitment risk.
- ↗To a standard per-token rate card: forgo the up-to-20%-off-list discount and pay as you go with no expiry.
Resources & guides
Tutorials & Learning
YouTube returned 6 videos for “Touchmark”, and we withheld 6: 6 could not be judged, because “Touchmark” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Touchmark.
Official links
What are the best Touchmark alternatives?
The closest alternatives to Touchmark we list are NannyML, Lmnr and ToolSpend. Ranked by direct product-type match, not category overlap.
| Alternative | Pricing | Starts at | Last verified |
|---|---|---|---|
| NannyML | Freemium | Pricing not published | — |
| Lmnr | Freemium | Pricing not published | — |
| ToolSpend | Paid | Pricing not published | — |
Alternatives to Touchmark
View all
Tools that pair well with Touchmark
Common stack mates teams adopt alongside Touchmark, with the specific reason each pairing earns its keep.
NannyML
NannyML estimates your ML model's live performance without ground-truth labels and alerts you only when it actually matters
Lmnr
Laminar delivers open-source AI agent observability that auto-captures every LLM and tool call and surfaces failures with Signals.
ToolSpend
Track AI spend across OpenAI, Anthropic, Google AI, Bedrock and Azure in one live cost dashboard.
Touchmark vs its closest rival
Touchmark's closest head-to-head is with Spider Cloud — If you need to feed live web data into AI agents or RAG pipelines, Spider Cloud is the clear choice with its Rust-powered crawling, AI Studio, and Browser AI commands. If your pain point is unpredictable AI API costs across multiple providers, Touchmark provides the transparency and budget controls you need. They solve completely different problems. Verdict generated
Head-to-head comparisons
Touchmark vs Spider Cloud
If you need to feed live web data into AI agents or RAG pipelines, Spider Cloud is the clear choice with its Rust-powered crawling, AI Studio, and Browser AI commands. If your pain point is unpredictable AI API costs across multiple providers, Touchmark provides the transparency and budget controls you need. They solve completely different problems.
GeneratedTouchmark vs Nectar Energy
Choose Nectar Energy if you manage commercial real estate HVAC/lighting and need AI-driven automation plus ESG reporting (now enhanced with GRESB/CDP). Choose Touchmark if you're a tech team struggling with unpredictable AI API costs and need granular per-model tracking. They serve entirely different domains — pick based on whether your pain is building energy or AI spend.
GeneratedTouchmark vs Temporal Ai
Buyer's choice depends entirely on pain point. If you need to build crash-resistant AI agents that survive failures and long waits, Temporal AI is the only durable execution platform with proven enterprise adoption. If your primary struggle is unpredictable AI API bills across multiple providers, Touchmark delivers granular cost visibility and alerts. They solve different problems — pick based on your biggest bottleneck.
GeneratedFrequently asked questions
Best-of guides
Used Touchmark? Help shape our editorial sentiment research.