AI infrastructure and model inference pricing
The layer between a foundation model and an application: vector databases for retrieval, agent frameworks for orchestration, and inference providers that serve open-weight models by the token. Pricing here is usage-based almost without exception, metered by tokens, queries or storage rather than sold as a flat monthly seat.
AI price index
How AI model prices are moving underneath these tools
$2
per million input tokens for the cheapest frontier LLM, 90% less than Nov 2023. Our view for 2028: $0.50 to $4, and $0.20 to $5 by 2030.
Open the AI price indexExhibit 1
Pinecone is the cheapest way in at $20, against $249 for Braintrust
Entry price of each product, USD per month
- PineconeBuilder$20
- LangfuseCore$29
- WeaviateFlex$45
- PortkeyProduction$49
- Arize AIAX Pro$50
- HeliconePro$79
- GalileoPro$100
- BraintrustPro$249
AI infrastructure and model inference compared
Current list prices as last read from each pricing page.
| Product | Starts at | Top listed plan | Priced by | Free plan | History |
|---|---|---|---|---|---|
 Arize AI | $50 per monthAX Pro, USD per month | contact sales | Flat per account | Yes | read Sep 2026 |
 AssemblyAI | $0.15 per hour of audioUniversal-3.5, USD per hour of audio | $0.21 per hour of audioUniversal-3.5 Pro | unknown | No | read Sep 2026 |
 Braintrust | $249 per month, $100 credits included, then usagePro, USD per month, platform fee | contact sales | Flat per account | Yes | read Sep 2026 |
 Cohere | $2500 per month per instance, or $4.00 per hourEmbed 4 Small, USD per month per instance | $3250 per month per instance, or $5.00 per hourRerank 3.5 Medium | Flat per account | No | read Sep 2026 |
 Deepgram | $0.0057 per minute of audioFlux Multilingual, USD per minute of audio | $0.0065 per minute of audioNova-3 | unknown | No | read Sep 2026 |
 Fireworks AI | $0.008 per million tokensServerless, up to 150M params, USD per million tokens | $0.10 per million tokensQwen3 8B | unknown | No | read Sep 2026 |
 Galileo | $100 per month, billed yearlyPro, USD per month, billed yearly | contact sales | Flat per account | Yes | read Sep 2026 |
 Groq | $0.05 per million input tokens, $0.08 per million outputLlama 3.1 8B Instant, USD per million input tokens | $0.59 per million input tokens, $0.79 per million outputLlama 3.3 70B Versatile | unknown | No | read Sep 2026 |
 Helicone | $79 per monthPro, USD per month | $799 per monthTeam | Flat per account | Yes | read Sep 2026 |
 Humanloop | contact sales | contact sales | Contact sales | No | read Sep 2026 |
 LangChain | $39 per seat per month, then pay as you goPlus, USD per seat per month, before usage | contact sales | Per seat | Yes | read Sep 2026 |
 Langfuse | $29 per monthCore, USD per month | $2499 per monthEnterprise | Flat per account | No | read Sep 2026 |
 LangWatch | €29 per core seat per month, 200K events includedGrowth, EUR per core seat per month | single paid plan | Per seat | Yes | read Sep 2026 |
 Modal | $0.000164 per second of computeNvidia T4, USD per second of compute | $0.001097 per second of computeNvidia H100 SXM5 | unknown | No | read Sep 2026 |
 OpenRouter | contact sales | contact sales | Usage based | Yes | read Oct 2026 |
 Pinecone | $20 per month, flat, solo developersBuilder, USD per month | $500 per month, minimum usageEnterprise | Flat per account | Yes | read Sep 2026 |
 Portkey | $49 per month, 100K logs includedProduction, USD per month | contact sales | Flat per account | Yes | read Nov 2026 |
 Replicate | $0.025 per output imageFLUX.1 schnell, USD per output image | $0.04 per output imageFLUX.1 dev | unknown | No | read Sep 2026 |
 Runpod | $0.89 per GPU hour, Secure Cloud on demand ($0.34 on Community Cloud)RTX 4090 pod, USD per GPU hour, Secure Cloud on-demand pod | $5.29 per GPU hour, Secure Cloud on demand ($3.59 on Community Cloud)H200 pod | Usage based | No | read Oct 2026 |
 Together AI | $0.35 per million input tokens, $1.50 per million outputMuse Glimmer 30B, USD per million input tokens | $3 per million input tokens, $15.00 per million outputKimi K3 | unknown | No | read Sep 2026 |
 Weaviate | $45 per month, starts at, pay-as-you-goFlex, USD per month, starting at | $400 per month, starts at, prepaid contractPremium | Flat per account | Yes | read Sep 2026 |
Head to head
Two products on one chart, plans side by side.
AI infrastructure and model inference pricing questions
- What does a AI infrastructure platform cost?
- Paid plans start at $50 per month for Arize AI (USD per month), $0.15 per hour of audio for AssemblyAI (USD per hour of audio), $249 per month, $100 credits included, then usage for Braintrust (USD per month, platform fee), $2500 per month per instance, or $4.00 per hour for Cohere (USD per month per instance), $0.0057 per minute of audio for Deepgram (USD per minute of audio), $0.008 per million tokens for Fireworks AI (USD per million tokens), $100 per month, billed yearly for Galileo (USD per month, billed yearly), $0.05 per million input tokens, $0.08 per million output for Groq (USD per million input tokens), $79 per month for Helicone (USD per month), $39 per seat per month, then pay as you go for LangChain (USD per seat per month, before usage), $29 per month for Langfuse (USD per month), €29 per core seat per month, 200K events included for LangWatch (EUR per core seat per month), $0.000164 per second of compute for Modal (USD per second of compute), $20 per month, flat, solo developers for Pinecone (USD per month), $49 per month, 100K logs included for Portkey (USD per month), $0.025 per output image for Replicate (USD per output image), $0.89 per GPU hour, Secure Cloud on demand ($0.34 on Community Cloud) for Runpod (USD per GPU hour, Secure Cloud on-demand pod), $0.35 per million input tokens, $1.50 per million output for Together AI (USD per million input tokens), $45 per month, starts at, pay-as-you-go for Weaviate (USD per month, starting at). Prices are as shown on each pricing page when it was last read.
- What is AI infrastructure platform pricing based on?
- Arize AI and Braintrust and Cohere and Galileo and Helicone and Langfuse and Pinecone and Portkey and Weaviate: flat per account. Humanloop: contact sales. LangChain and LangWatch: per seat. OpenRouter and Runpod: usage based.
- Which ai infrastructure and model inference have a free plan?
- Arize AI, Braintrust, Galileo, Helicone, LangChain, LangWatch, OpenRouter, Pinecone, Portkey, Weaviate list a free plan next to the paid tiers.
Other categories
- Password managers
- Project management tools
- Team communication
- AI assistants
- AI coding tools
- AI media generation
- AI meeting notetakers
- AI voice agent platforms
- AI search visibility tools
- AI agent builders
- SQL editors and database tools
- Data analysis and BI tools
- Automation and marketing tools
- CRM software
- Office suites and business email
- Sales intelligence and B2B data
- Design tools
- Developer platforms
- E-commerce platforms
- Web hosting
- Music streaming
- Video streaming and live TV
- Gaming subscriptions
- Audiobooks and ebooks
- Customer support software
- Cloud storage
- Email marketing
- Notes and docs
- SEO tools
- AI model APIs
- Monitoring and observability
- CI/CD and testing tools
- Identity and security tools
- Billing and payments
- Accounting and expense tools
- Product analytics and onboarding
- E-signature and document tools
- Video and webinar tools
- HR and recruiting software
- Communication APIs and telephony
- Integration and automation platforms
- Managed databases and search
- Surveys and customer feedback
- Applicant tracking systems
- Job boards
- Security compliance automation
- Security awareness training
- Sales engagement and revenue intelligence
- Enterprise AI agents
- Customer success platforms
- Employee engagement and performance
- Content moderation
- Social media management
- Employee scheduling software
- Learning management and course platforms
- Product management and roadmapping
- Localization and translation management
- Time tracking software
- Remote access and support
- Event management platforms
- Intranet and internal communications
- Construction management software
- Field service management
- IT service management and help desk
- Appointment booking software
- Privacy-friendly web analytics
- Live chat and chatbots
- Web scraping
- MCP gateways