Forge is preparing the requested surface and verifying the live route.
Forge is preparing the requested surface and verifying the live route.
Forge subscriptions bundle the full managed engine layer. Tools still support standalone subscriptions, and tools plus engines can expose direct metered access for outside agents while enterprise can enable the full catalog and pay only for usage.
These plans are for customers who want the full managed Forge platform. Every paid plan includes the managed engine layer, so subscribed users do not have to buy each engine separately.
$0
1 agent · community usage limits
$50/mo
900 req/hr · 8,640 req/day · 10 parallel
$100/mo
2,700 req/hr · 25,920 req/day · 30 parallel
$200/mo
8,100 req/hr · 77,760 req/day · 90 parallel
$400/mo
24,300 req/hr · 233,280 req/day · 270 parallel
Published rate card
Full catalog enabled · pay only for usage · automatic 0-50% bulk discounts
Enterprise is not unpublished custom pricing. The unified pricing model uses the same published per-module USD rate card with automatic discounts based on trailing 30-day spend, plus invoice terms and prepaid options when procurement needs them. Enterprise can enable the full Forge catalog at once and only gets billed for the surfaces that actually run.
Trailing 30-day spend
$0+
0% off
Standard
Trailing 30-day spend
$10k+
5% off
Commercial
Trailing 30-day spend
$50k+
10% off
Business
Trailing 30-day spend
$100k+
15% off
Scale
Trailing 30-day spend
$250k+
25% off
Enterprise
Trailing 30-day spend
$500k+
35% off
Strategic
Trailing 30-day spend
$1M+
50% off
Maximum
Forge reduces duplicated compute through cache hits, prefix reuse, ICE compression, memory reuse, and routing intelligence. Bundled subscribers get those savings across the full platform, and metered buyers still benefit from the same underlying optimization layer.
Forge's caching, compression, and routing make your LLM calls cheaper than going direct — even with our fees included.
Forge records every dollar saved on every call. These numbers are live — not estimates.
See how it worksCards and ACH stay available for conventional checkout. Native x402 and Stripe MPP are the machine-payment rails that make direct tool and engine access work for autonomous systems.
Forge subscriptions and selected checkout flows use Stripe card and ACH rails where appropriate.
Autonomous agents and outside systems can pay for tools and engines over native machine-payment rails.
Stripe machine payments stay available as a separate programmable rail for automated settlement.
The locked Forge individual subscription plans stay unchanged. Everything below covers standalone tools, module commits, engine pricing, governed formulas, commercial systems, and the enterprise full-catalog usage model that sits outside the bundled platform plans.
Enterprise can turn on the full Forge catalog and pay only for the tools, modules, engines, agents, and model usage that actually run. Automatic bulk discounts apply by published rate-card category.
| Surface | Published pricing | Enterprise treatment |
|---|---|---|
| Full catalog access | Enable every published Forge surface and pay only for metered usage | Enterprise buyers do not need to pre-select tools, modules, engines, or agents. The catalog can be enabled up front and billing only tracks what gets used. |
| Agents and autonomous execution | Simple $0.003/step · Standard $0.008/step · Complex $0.020/step | Agent execution, crews, and autonomous workflows all inherit automatic bulk discounts. Machine-payment settlement remains available through x402 and Stripe MPP. |
| Hosted models, BYOK, and provider routing | Inference Nano $0.02/$0.04 · Small $0.05/$0.10 · Medium $0.10/$0.20 · Large $0.20/$0.50 · Frontier $0.40/$1.00 per 1M input/output tokens · BYOK platform fee $0.10 per 1M tokens | Enterprise can route across the full Forge model catalog and BYOK providers without separate product approvals. Pricing stays usage-based and discounts auto-apply by category. |
| HF imports, VMFS/HDT-style model delivery, and Nucleus-managed models | Model download $0.10/GB · Storage $0.020/GB-month · Nucleus inference surcharge $0.005/1M tokens · Hot slot 20% of PVL tier with $5 minimum · PVL tiers from $25 to $10,000 per model-month | Hosted OSS models, Hugging Face imports, compressed model distribution, hot residency, and managed-model value capture all use the published Nucleus rate card with automatic bulk discounts. |
| GPU, voice, video, and code execution | Published GPU and code-execution rate card with category discounts up to 50% | Image, video, TTS, 3D, and managed execution workloads are enterprise-eligible without custom contracts. Customers pay only for actual runtime usage. |
| Standalone tools, packaged engines, and frontier systems | Published standalone subscriptions plus PAYG for tools and engines, with enterprise usage billed against the same public rate card | Enterprise can expose the full catalog for teams and agents, then let usage determine spend instead of negotiating product-by-product entitlements. |
| Commercial systems and royalty surfaces | Published take rates, usage formulas, and royalty policies with enterprise bulk treatment where module billing applies | Commercial systems stay price-transparent. Back-office settlement and reconciliation controls remain internal even when the customer-facing economics are public. |
Tools sell outcomes. Each one gets pay-per-use access plus recurring subscriptions with included usage and overage.
| Tool | Value metric | Pay per use | Starter | Team | Scale |
|---|---|---|---|---|---|
| ICE | 1M source tokens compressed | $0.30 | $50/mo · 100M tokens | $200/mo · 1B tokens | $400/mo · 5B tokens |
| ForgeGuard | 1k guarded requests | $0.75 | $100/mo · 250k requests | $200/mo · 2M requests | $400/mo · 15M requests |
| Forge Memory | 100k memory ops | $0.20 | $50/mo · 5M ops | $200/mo · 25M ops | $400/mo · 150M ops |
| Forge Cache | 100k cache ops | $0.10 | $50/mo · 10M ops | $100/mo · 75M ops | $400/mo · 500M ops |
| Forge Router | 100k routed requests | $0.25 | $50/mo · 2M requests | $200/mo · 15M requests | $400/mo · 100M requests |
| Forge Audit | 10k audit events | $2.50 | $100/mo · 500k events | $200/mo · 5M events | $400/mo · 40M events |
| Forge Search | 1k research credits | $12.00 | $50/mo · 5k credits | $200/mo · 30k credits | $400/mo · 150k credits |
| Forge Observe | 100k observability units | $10.00 | $50/mo · 500k units | $200/mo · 5M units | $400/mo · 40M units |
Bundle pricing for buyers who want multiple standalone Forge tools without buying the full Forge platform subscription.
| Bundle | Price | Includes |
|---|---|---|
| Tools Pro Bundle | $200/mo | all 8 tools, 1 workspace, pooled overage |
| Tools Team Bundle | $400/mo | all 8 tools, 10 seats, pooled overage |
Modules remain metered primitives. Their subscriptions are recurring usage commits, not unlimited flat-access plans.
| Commit tier | Monthly commit | Discount vs PAYG | Best for |
|---|---|---|---|
| Developer Commit | $50/mo | 5% | trying one module in production |
| Team Commit | $250/mo | 10% | stable recurring use |
| Growth Commit | $1,000/mo | 15% | multiple workloads on one module |
| Scale Commit | $5,000/mo | 25% | sustained multi-team volume |
| Strategic Commit | $10,000+/mo | 35% to 50% | very large recurring usage |
Published PAYG module rates. Enterprise and API buyers use this rate card with automatic bulk discounts.
| Module | Unit | Published rate |
|---|---|---|
| faer | per routing decision | $0.0001 |
| inference_nano | per 1M input/output tokens | $0.02 in / $0.04 out |
| inference_small | per 1M input/output tokens | $0.05 in / $0.10 out |
| inference_medium | per 1M input/output tokens | $0.10 in / $0.20 out |
| inference_large | per 1M input/output tokens | $0.20 in / $0.50 out |
| inference_frontier | per 1M input/output tokens | $0.40 in / $1.00 out |
| byok_platform_fee | per 1M tokens | $0.10 |
Published PAYG module rates. Enterprise and API buyers use this rate card with automatic bulk discounts.
| Module | Unit | Published rate |
|---|---|---|
| agent_step_simple | per step | $0.003 |
| agent_step_standard | per step | $0.008 |
| agent_step_complex | per step | $0.020 |
| crystal_engine | per pass | $0.0005 |
| fcse_query | per query | $0.0002 |
| fcse_chain_build | per chain | $0.001 |
| cle_fact_query | per query | $0.0001 |
| cle_fact_write | per write | $0.0003 |
| fmf_operation | per operation | $0.0003 |
| forgehive_routing | per routing event | $0.00005 |
| fragment_assembly | per assembly | $0.0002 |
Published PAYG module rates. Enterprise and API buyers use this rate card with automatic bulk discounts.
| Module | Unit | Published rate |
|---|---|---|
| citadel_threat_genome | per check | $0.001 |
| citadel_behavioral_crypto | per entity update | $0.005 |
| citadel_zk_proofs | per proof | $0.020 + $0.050 anchor |
| citadel_canary_swarm | per canary event | $0.0005 |
| citadel_chain_of_custody | per tag | $0.0002 |
| citadel_provider_integrity | per probe | $0.010 |
| citadel_swat | per trigger | $0.50 |
| citadel_consensus | per round | $0.005 |
Published PAYG module rates. Enterprise and API buyers use this rate card with automatic bulk discounts.
| Module | Unit | Published rate |
|---|---|---|
| gpu_image_512 | per image | $0.003 |
| gpu_image_1024 | per image | $0.010 |
| gpu_image_2048 | per image | $0.030 |
| gpu_image_edit | per operation | $0.008 |
| gpu_video | per second | $0.080 |
| gpu_tts | per minute | $0.015 |
| gpu_3d | per asset | $0.250 |
| code_execution_standard | per minute | $0.005 |
| code_execution_performance | per minute | $0.018 |
| code_execution_heavy | per minute | $0.040 |
Published PAYG module rates. Enterprise and API buyers use this rate card with automatic bulk discounts.
| Module | Unit | Published rate |
|---|---|---|
| forge_lens | per session | $0.050 |
| forge_token | per 1k tracked tokens | $0.001 |
| forge_sentinel | per 1k screened requests | $0.002 |
| forge_pulse | per cycle | $0.010 |
| forge_checkpoint | per event | $0.001 |
| forge_vault | per access | $0.002 |
| forge_compress | per operation | $0.0003 |
| forge_iq | per insight | $0.030 |
| forge_consensus | per round | $0.005 |
| forge_sync | per GB | $0.020 |
| forge_optimizer | per cycle | $0.020 |
| forge_mirror | per GB | $0.015 |
| forge_predict | per prediction | $0.005 |
| forge_instant | per cached response | $0.0001 |
Published PAYG module rates. Enterprise and API buyers use this rate card with automatic bulk discounts.
| Module | Unit | Published rate |
|---|---|---|
| overage_starter | per request | $0.001 |
| overage_pro | per request | $0.0004 |
| overage_growth | per request | $0.0002 |
| overage_ultimate | per request | $0.0001 |
| extended_context_32k | per request per 1k tokens | $0.0005 |
| extended_context_128k | per request per 1k tokens | $0.0015 |
Published PAYG module rates. Enterprise and API buyers use this rate card with automatic bulk discounts.
| Module | Unit | Published rate |
|---|---|---|
| nucleus_model_download | per GB downloaded | $0.10 |
| nucleus_storage | per GB-month | $0.020 |
| nucleus_inference_call | per 1M tokens | $0.005 |
| nucleus_hot_slot | per hot slot-month | 20% of PVL tier, $5 minimum |
| nucleus_pvl_fee | per managed model-month | tiered by monthly call volume |
| forge_compress_pvl_fee | per GB-month | $0.005 |
| forge_mirror_pvl_fee | per GB-month | $0.010 |
| forge_vault_pvl_fee | per 1k secrets-month | $1.00 |
| forge_sync_pvl_fee | per channel-month | $3.00 |
| forge_optimizer_pvl_fee | per pipeline-month | $5.00 |
| ipre_royalty | percent of attributable net profit | default 2% |
| ipre_discovery_session | per tracked session | $0.50 |
Core engines package orchestration and managed control planes on top of primitive module billing.
| Engine | Primary metric | Pay per use | Starter | Team | Scale |
|---|---|---|---|---|---|
| Forge Memory | 100k memory ops | $0.20 | $50/mo | $200/mo | $400/mo |
| ForgeGuard | 1k guarded requests | $0.75 | $100/mo | $200/mo | $400/mo |
| FMF | 100k fabric ops | $0.15 | $50/mo | $100/mo | $400/mo |
| ICE | 1M compressed tokens | $0.30 | $50/mo | $200/mo | $400/mo |
| CLE | 100k reads / writes | $0.25 reads / $0.60 writes | $50/mo | $200/mo | $400/mo |
| Healing | 1k incident events | $0.40 | $50/mo | $200/mo | $400/mo |
| Validation | 1k validations | $0.30 | $50/mo | $100/mo | $400/mo |
| Atlas | 1k knowledge ops | $0.35 | $100/mo | $200/mo | $400/mo |
| Weave | 1k mesh ops | $0.40 | $100/mo | $200/mo | $400/mo |
Frontier engines are priced by decision quality, strategic leverage, and economic upside. Higher-value systems move away from flat pricing and toward value-linked usage.
| Engine | Primary metric | Pay per use | Starter | Team | Scale |
|---|---|---|---|---|---|
| Liquid Weights | 100k optimized routes | $0.20 | $499/mo | $1,999/mo | $7,500/mo |
| Entangled Inference | 1M coordinated tokens | $3.00 | $750/mo | $2,500/mo | $10,000/mo |
| Semantic Compilation | 1M compiled tokens | $1.50 | $399/mo | $1,499/mo | $5,000/mo |
| Zero-Weight Inference | 1M synthetic retrieval tokens | $2.00 | $299/mo | $999/mo | $3,500/mo |
| Temporal Inference | per analysis | $0.02 | $249/mo | $999/mo | $3,500/mo |
| SIC | per substrate compile | $0.50 | $499/mo | $1,999/mo | $7,500/mo |
| LKS | per knowledge evolution op | $0.01 | $299/mo | $1,199/mo | $4,000/mo |
| SEE | per state enrichment | $0.008 | $249/mo | $999/mo | $3,500/mo |
| Causal Memory | 100k causal graph ops | $0.30 | $299/mo | $1,199/mo | $4,000/mo |
| ECI | per debate round | $0.08 | $599/mo | $2,000/mo | $8,000/mo |
| Civilization Fork | per fork-hour | $25.00 | $2,500/mo | $10,000/mo | custom bulk, published formula |
| TA | per goal check | $0.004 | $299/mo | $1,199/mo | $4,000/mo |
| Proactive IE | per scored event | $0.006 | $199/mo | $799/mo | $2,999/mo |
| Collective Mesh | per mesh signal | $0.01 | $349/mo | $1,399/mo | $5,000/mo |
| World Simulator | per run or sim-minute | $2.00/run or $0.25/sim-minute | $1,500/mo | $6,000/mo | $20,000/mo |
| Epistemic Integrity | per claim check | $0.006 | $399/mo | $1,499/mo | $5,000/mo |
| Autonomous Research | per investigation kickoff | $25.00 plus usage | $1,500/mo | $6,000/mo | $20,000/mo |
Nucleus Ultra is not sold as a static one-price engine. It is packaged around managed-model count, PVL tier, storage, and hot-slot value.
| Package | Price | Includes |
|---|---|---|
| Nucleus Ultra Launch | $299/mo | 2 managed models in Low PVL tier, 50 GB compressed storage, 1 hot slot |
| Nucleus Ultra Growth | $1,500/mo | 10 managed models up to Active tier, 500 GB storage, 5 hot slots |
| Nucleus Ultra Scale | $7,500/mo | 50 managed models up to Commercial tier, 5 TB storage, 25 hot slots |
Commercial moonshot systems need explicit economics too. These are published pricing formulas, not hidden custom contract logic.
| System | Pricing | Notes |
|---|---|---|
| Genome Market | 10% marketplace take rate | drops to 5% for sellers on a $99/mo seller plan |
| Causal Commerce | 1.5% attributable revenue fee | plus $0.10 per claim and $1.00 per settlement |
| Morphic Commerce | 8% take rate on pack sales | plus $0.50 per paid injection |
| Temporal Market | $25 per quote pack | plus 5% take rate on accepted temporal bids |
| Adversarial Economy | $499/mo control plane | plus 10% of bounty payouts processed |
Some surfaces should have published formulas but not raw unrestricted self-serve primitive access. These remain governed buying paths.
| Surface | Published formula |
|---|---|
| Nucleus PVL | Dormant $0, Low $25, Active $150, Commercial $750, Scale $3,000, Hyperscale $10,000 |
| Nucleus Hot Slot | 20% of active PVL tier with $5/month minimum |
| IP Royalty Engine | default 2% of attributable net profit, bounded by published policy |
Everything that creates customer value can be sold, but not every underlying control-plane route should become a public endpoint.
| Policy area | Internal-only rule |
|---|---|
| Operator-only raw moonshots | Causal Memory and Civilization Fork raw routes should stay internal even when their packaged engines are sold. |
| Settlement and reconciliation controls | Commercial moonshot settlement, reconciliation, reward, and controller endpoints should remain internal-only back-office surfaces. |
| Training and mutation maintenance | Engine training, flush, ontology-apply, and specialist retraining endpoints should not be exposed as public payable primitives. |
| Subscription overage artifacts | Overage modules are billing internals for the locked Forge plans, not standalone purchasable products. |
Humans and autonomous agents use the same catalog. Pay once with card, bank, native x402, or Stripe MPP, or create a persistent subscription with a scoped API key.
No public tools are published right now
Forge is not currently publishing any pay-per-use tools in the public catalog. Check back later or create an account for workspace access.
Machine quote network
Public pay-per-use quotes now respect the network you choose instead of assuming a single default rail.
No subscription required
Quote and pay for a single call from the public tool catalog.
External-agent ready
The same pay-per-use rails work for bots, scripts, and any outside platform.
Subscription available
Create an account when you want scoped API keys and persistent tool subscriptions.
Paid Forge subscribers get the full engine layer bundled in. Non-subscribers, autonomous agents, and enterprise buyers can still access individual engines through published pay-per-use rails.
Machine quote network