Cernio — Pricing & Cost Analysis
Last updated: 2026-03-31 Status: DRAFT — to be finalized before beta Reference: docs/strategy/04-billing-credits.md
1. AI API Model Pricing (March 2026)
1.1 Google Gemini
| Model | Input $/1M | Output $/1M | Context | Web Search (Grounding) | Batch | Notes |
|---|---|---|---|---|---|---|
| Gemini 2.5 Flash-Lite | $0.10 | $0.40 | 1M | $35/1K queries | 50% | Cheapest Gemini, ideal for batch/classify |
| Gemini 2.5 Flash | $0.30 | $2.50 | 1M | $35/1K queries | 50% | Project default (quick tier) |
| Gemini 2.5 Pro (<=200K) | $1.25 | $10.00 | 1M | $35/1K queries | 50% | For deep analysis |
| Gemini 2.5 Pro (>200K) | $2.50 | $15.00 | 1M | $35/1K queries | 50% | Long context |
| Gemini 3 Flash (Preview) | TBD | TBD | 1M | $14/1K queries | — | Grounding 2.5x cheaper! Preview |
| Gemini 3 Pro (<=200K, Preview) | $2.00 | $12.00 | 1M | $14/1K queries | — | Preview, pricing may change |
| Gemini 3 Pro (>200K, Preview) | $4.00 | $18.00 | 1M | $14/1K queries | — | Preview |
- Content fetched via Grounding is not billed as input tokens
- Cache reads: 10% of base input price
- Batch API: 50% discount on all models, 24-hour turnaround
1.2 OpenAI
| Model | Input $/1M | Output $/1M | Context | Web Search | Batch | Notes |
|---|---|---|---|---|---|---|
| GPT-4.1 Nano | $0.10 | $0.40 | 1M | $25/1K + 8K token/call | 50% | Cheapest GPT, ideal for classify |
| GPT-4.1 Mini | $0.40 | $1.60 | 1M | $25/1K + 8K token/call | 50% | Good balance |
| GPT-4.1 | $2.00 | $8.00 | 1M | $30/1K + token | 50% | GPT-4o successor |
| GPT-4o Mini | $0.15 | $0.60 | 128K | $25/1K + 8K token/call | 50% | Legacy, prefer 4.1 Mini |
| GPT-4o | $2.50 | $10.00 | 128K | $30/1K + token | 50% | Project balanced tier |
| GPT-5 | $1.25 | $10.00 | — | — | — | New flagship |
| o3 | $2.00 | $8.00 | 200K | — | — | Reasoning |
| o3-mini | $1.10 | $4.40 | 200K | — | 50% | Reasoning (cheap) |
| o4-mini | $1.10 | $4.40 | 200K | — | 50% | Reasoning (new) |
- GPT-4.1 series is GPT-4o’s successor, 1M context, same or cheaper
- Web search: fixed 8K input token block + per-call fee (cost higher than expected)
- o-series models are for reasoning, generally unnecessary for lead gen
1.3 Anthropic (Claude)
| Model | Input $/1M | Output $/1M | Context | Web Search | Batch | Notes |
|---|---|---|---|---|---|---|
| Claude Haiku 4.5 | $1.00 | $5.00 | 200K | None | 50% | Fast, medium quality |
| Claude Sonnet 4.6 | $3.00 | $15.00 | 200K | None | 50% | Project deep tier |
| Claude Opus 4.6 | $15.00 | $75.00 | 200K | None | 50% | Most powerful, very expensive |
- Prompt caching: 90% discount (cache hit)
- No web search support — not suitable alone for headhunt/discovery
- Extended thinking: Sonnet 60/M (input)
1.4 Perplexity
| Model | Input $/1M | Output $/1M | Request Fee /1K | Context | Notes |
|---|---|---|---|---|---|
| Sonar (Low) | $1.00 | $1.00 | $5 | 128K | Cheapest web search |
| Sonar (Medium) | $1.00 | $1.00 | $8 | 128K | Balanced |
| Sonar (High) | $1.00 | $1.00 | $12 | 128K | Comprehensive search |
| Sonar Pro (Low) | $3.00 | $15.00 | $6 | 200K | Pro quality, cheap search |
| Sonar Pro (Medium) | $3.00 | $15.00 | $10 | 200K | Project headhunt tier |
| Sonar Pro (High) | $3.00 | $15.00 | $14 | 200K | Most comprehensive |
| Sonar Deep Research | $2.00 | $8.00 | Variable | — | Deep research |
- Citation tokens are not billed (Sonar + Sonar Pro)
- Request fee + token fee are calculated together
- Ideal for headhunt (web search included)
- Example: Sonar Low, 500 input + 200 output = ~$0.006/query
1.5 Budget Alternatives
| Provider | Model | Input $/1M | Output $/1M | Context | Web Search | Notes |
|---|---|---|---|---|---|---|
| DeepSeek | V4 | $0.30 | $0.50 | — | None | Cache hit: $0.03 input |
| DeepSeek | V3.2 | $0.28 | $0.42 | — | None | Cache hit: $0.028 |
| DeepSeek | R1 (reasoning) | $0.55 | $2.19 | — | None | Reasoning model |
| Mistral | Ministral-8B | $0.10 | $0.10 | — | None | Cheapest Mistral |
| Mistral | Medium 3 | $0.40 | $2.00 | — | None | Mid-range |
| Mistral | Large 3 | $2.00 | $6.00 | — | None | Most powerful Mistral |
| Groq | Llama 3.1 8B | $0.05 | $0.08 | 128K | None | Ultra fast, ultra cheap |
| Groq | Llama 3.3 70B | $0.59 | $0.79 | 128K | None | Batch 50% discount |
1.6 Price Comparison Summary (Cheapest → Most Expensive)
| Segment | Model | Input | Output | Use Case |
|---|---|---|---|---|
| Ultra Cheap | Groq Llama 3.1 8B | $0.05 | $0.08 | Simple classify |
| Cheap | Gemini 2.5 Flash-Lite / GPT-4.1 Nano | $0.10 | $0.40 | Batch classify, simple analysis |
| Economical | Gemini 2.5 Flash / GPT-4.1 Mini | $0.30-0.40 | $1.60-2.50 | Discovery, enrichment |
| Balanced | GPT-4.1 / GPT-4o / Gemini 2.5 Pro | $1.25-2.50 | $8-10 | Deep analysis, enrichment |
| Premium | Claude Sonnet 4.6 | $3.00 | $15.00 | Highest quality output |
| Ultra Premium | Claude Opus 4.6 | $15.00 | $75.00 | Rarely needed |
| Web Search | Perplexity Sonar (Low) | $1.00 | 5/1K req | Headhunt, contact finding |
2. Cernio Operational Costs (Per Model)
2.1 Token Usage (Measured from Project)
| Operation | System Prompt | User Input | Output | Total Tokens |
|---|---|---|---|---|
| Discovery (1 search, 15 companies) | ~300 | ~700 | ~2,500 | ~3,500 |
| Headhunt (1 company, 3 contacts) | ~200 | ~700 | ~800 | ~1,700 |
| Enrichment (1 company analysis) | ~300 | ~1,300 | ~1,700 | ~3,300 |
| Batch classify (25 companies) | ~250 | ~1,500 | ~600 | ~2,350 |
2.2 Discovery Cost — Model Comparison
1 discovery search = ~1,000 input + ~2,500 output tokens + web search| Model | Token Cost | Web Search | Total /search |
|---|---|---|---|
| Gemini 2.5 Flash-Lite | $0.0011 | $0.035 | $0.036 |
| Gemini 2.5 Flash | $0.0010 | $0.035 | $0.036 |
| Gemini 2.5 Pro | $0.026 | $0.035 | $0.061 |
| Gemini 3 Flash (Preview) | TBD | $0.014 | ~$0.015 |
| GPT-4.1 Nano | $0.0011 | $0.025 + token | ~$0.027 |
| GPT-4.1 Mini | $0.0044 | $0.025 + token | ~$0.030 |
| GPT-4.1 | $0.022 | $0.030 + token | ~$0.053 |
| Perplexity Sonar (Low) | $0.0035 | Included ($0.005/req) | $0.009 |
| Perplexity Sonar Pro (Low) | $0.041 | Included ($0.006/req) | $0.047 |
2.3 Headhunt Cost — Model Comparison
1 headhunt = ~900 input + ~800 output tokens + web search (REQUIRED)| Model | Token Cost | Web Search | Total /contact |
|---|---|---|---|
| Perplexity Sonar (Low) | $0.0017 | Included ($0.005/req) | $0.007 |
| Perplexity Sonar (Medium) | $0.0017 | Included ($0.008/req) | $0.010 |
| Perplexity Sonar Pro (Low) | $0.015 | Included ($0.006/req) | $0.021 |
| Perplexity Sonar Pro (Medium) | $0.015 | Included ($0.010/req) | $0.025 |
| Gemini 2.5 Flash + Grounding | $0.0022 | $0.035 | $0.037 |
| Gemini 3 Flash + Grounding | TBD | $0.014 | ~$0.015 |
| GPT-4.1 Mini + Search | $0.0016 | $0.025 + 8K token | ~$0.030 |
2.4 Enrichment Cost — Model Comparison
1 enrichment = ~1,600 input + ~1,700 output tokens (web search optional)| Model | Total /company | Notes |
|---|---|---|
| Gemini 2.5 Flash-Lite | $0.0008 | Quality needs testing |
| Gemini 2.5 Flash | $0.005 | Good balance |
| GPT-4.1 Nano | $0.0008 | Cheap but quality? |
| GPT-4.1 Mini | $0.003 | Good balance |
| GPT-4.1 | $0.017 | Deep analysis |
| GPT-4o | $0.021 | Currently used in project |
| Claude Haiku 4.5 | $0.010 | Fast, medium quality |
| Claude Sonnet 4.6 | $0.030 | Highest quality |
| DeepSeek V4 | $0.001 | Budget alternative |
2.5 Batch Classify Cost (25 companies/batch)
| Model | Total /batch | /company |
|---|---|---|
| Groq Llama 3.1 8B | $0.0001 | $0.000004 |
| Gemini 2.5 Flash-Lite | $0.0004 | $0.000016 |
| GPT-4.1 Nano | $0.0004 | $0.000016 |
| Gemini 2.5 Flash | $0.002 | $0.00008 |
| GPT-4.1 Mini | $0.002 | $0.00008 |
3. Scenario-Based Cost Table
3.1 Pro Plan User — Typical Monthly Usage
50 discovery + 100 headhunt + 20 enrichment| Scenario | Discovery Model | Headhunt Model | Enrichment Model | Total/month |
|---|---|---|---|---|
| Cheap (minimum) | Sonar Low | Sonar Low | Flash-Lite | 0.70 + 1.17 |
| Balanced (recommended) | Gemini Flash | Sonar Low | GPT-4.1 Mini | 0.70 + 2.56 |
| Quality (current) | Gemini Flash | Sonar Pro | GPT-4o | 2.10 + 4.32 |
| Premium | GPT-4.1 | Sonar Pro Med | Claude Sonnet | 2.50 + 5.75 |
- Cheap scenario: 1.17 = $47.83 (97.6% margin)
- Balanced scenario: 2.56 = $46.44 (94.8% margin)
- Quality scenario: 4.32 = $44.68 (91.2% margin)
- Premium scenario: 5.75 = $43.25 (88.3% margin)
3.2 Team Plan User — Typical Monthly Usage
200 discovery + 500 headhunt + 100 enrichment| Scenario | Total/month | Margin ($149) |
|---|---|---|
| Cheap | 3.50 + 5.38** | $143.62 (96.4%) |
| Balanced | 3.50 + 11.00** | $138.00 (92.6%) |
| Quality | 10.50 + 19.80** | $129.20 (86.7%) |
| Premium | 12.50 + 26.10** | $122.90 (82.5%) |
3.3 Free Plan User — Monthly Cost (Cernio pays)
5 discovery + 5 headhunt + 0 enrichment (beta)| Scenario | Total/month |
|---|---|
| Cheap | $0.08 |
| Balanced | $0.22 |
4. Competitor Analysis (Detailed)
4.1 Tier 1 — Direct Competitors
Apollo.io
- Free: $0 — 100 credits/mo (email = 1cr, phone = ~8cr)
- Basic: $49/user/mo (annual) — 5,000 credits/mo
- Professional: $79/user/mo — 10,000 credits/mo
- Organization: $119/user/mo — 15,000 credits/mo
- Extra credits: $0.20/credit
- Credit rollover: NONE
- API: All plans (limited)
- AI: Email writing, lead scoring (Pro+), enrichment
- Commitment: None (monthly cancellation)
- 275M+ contact DB — significant gap vs Cernio
Lusha
- Free: $0 — 70 credits/mo
- Pro: $22/user/mo (annual) — 3,000 credits/year (~250/mo)
- Premium: $52/user/mo (annual) — 7,200 credits/year (~600/mo)
- Scale: Custom (~$15K/year)
- Credits: Email = 1cr, phone = 5cr
- Rollover: Accumulates up to 2x monthly
- API: Premium+
- Commitment: None
- Lowest entry point ($22/mo)
Seamless.AI
- Free: 50 credits (one-time, non-renewing)
- Basic: $147/mo (annual) — 250 credits/mo
- Pro: Custom (~$79-299/user/mo) — ~1,000 credits/day
- Enterprise: $20K-100K+/year
- Rollover: NONE — resets daily (on Pro!)
- API: Enterprise only
- AI: Real-time contact search
- Commitment: Pro requires min 5 users
- WARNING: Opaque pricing, contract traps reported
Cognism
- Grow: ~15-25K platform
- Elevate: ~$2,500/user/year + platform
- Typical: $15,000-40,000/year
- Credits: “Unlimited” (fair-use limit applies)
- API: All plans (limited)
- AI: Intent signals, phone-verified mobile
- Commitment: 12 months
- Europe-focused, GDPR compliant — high entry barrier
RocketReach
- Free: Limited
- Essentials: $33/mo (annual) — 1,200 email lookups/year
- Pro: $75/mo (annual) — 3,600 lookups/year (phone included)
- Ultimate: $175/mo (annual) — 10,000 lookups/year + API
- Extra credits: $0.30-0.45/lookup
- WARNING: Dual-credit system (lookup vs export are SEPARATE)
4.2 Tier 2 — Major Players
ZoomInfo
- Professional: ~$15K/year (3-5 users)
- Advanced: ~$25K/year
- Elite: ~$40K/year
- Enterprise (25+): $80-150K/year
- Credits: Starter 2,500/year (insufficient)
- Commitment: 12 months, difficult exit process
LinkedIn Sales Navigator
- Core: 100 (monthly)
- Advanced: $125/mo (annual)
- Advanced Plus: ~$1,600/user/year
- No data export — LinkedIn only
Clearbit (HubSpot Breeze Intelligence)
- 100 credits/mo: 30/mo min)
- 1,000 credits/mo: $150/mo
- 10,000 credits/mo: $700/mo
- Credit cost: ~$0.07-0.10/record
- Standalone API DISCONTINUED — HubSpot ecosystem only
4.3 Tier 3 — AI-First New Generation
Clay
- Free: $0 — 100 data credits + 500 actions/mo
- Launch: $185/mo — 2,500 data credits + 15K actions
- Growth: $495/mo — 6,000 data credits + 40K actions
- Enterprise: Custom
- Credit cost: Basic: 6-20 cr/record, full enrichment: ~75 cr
- 75+ data provider integrations — strongest enrichment
Persana.ai
- Free: $0 — 50 emails + 5 phone numbers
- Starter: $68/mo (annual) — 2,000 credits + 200 phone
- Growth: $151/mo (annual) — 5,000 credits + 500 phone
- Pro: $400/mo — 18,000 credits
- Unlimited: $600/mo — 50,000 credits
- Credit rollover: YES — accumulation supported
- AI lookalike, job tracking, AI SDR
Instantly.ai
- Outreach Growth: $30/mo — unlimited email accounts
- Lead DB Growth: $47/mo — 1,500-2,000 credits
- Real cost: $77-175/mo (outreach + leads + domain)
- Primary product: Email outreach, lead gen is secondary
Lemlist
- Email Pro: $55/user/mo (annual)
- Multichannel: $83/user/mo (annual)
- Primary product: Outreach sequencing, lead finding is secondary
4.4 Competitor Comparison Matrix
| Competitor | Entry Price | Credits/mo (entry) | AI Discovery | AI Headhunt | Web Search | Credit Rollover |
|---|---|---|---|---|---|---|
| Apollo.io | $49 | 5,000 | Limited | Yes | No | None |
| Lusha | $22 | ~250 | No | Yes | No | 2x |
| Seamless.AI | $147 | 250 | No | Real-time | Yes | None |
| Cognism | $1,250+ | “Unlimited” | No | Yes | Intent | N/A |
| RocketReach | $33 | 100 | No | Yes | No | None |
| ZoomInfo | $1,250+ | Custom | Intent | Yes | Intent | None |
| Clay | $185 | 2,500 | Yes (AI) | Yes (AI) | Yes (75+) | Unclear |
| Persana.ai | $68 | 2,000 | Lookalike | Yes (AI) | Yes | Yes |
| Instantly.ai | $77 | 1,500 | Limited | DB | No | Unclear |
| Cernio | $49 | 50 disc + 100 reveal | Yes (AI) | Yes (AI) | Yes | Yes (2x) |
5. Cernio Pricing Decisions
5.1 Plan Prices (Finalized)
| Plan | Monthly | Annual (-20%) | Notes |
|---|---|---|---|
| Free | $0 | — | For WOW moment, unlimited duration |
| Pro | $49/mo | 468/year) | Same tier as Apollo Basic |
| Team | $149/mo | 1,428/year) | Cheaper than Clay Launch |
| Enterprise | Custom | Custom | Sales conversation |
5.2 Plan Limits
| Feature | Free | Pro | Team | Enterprise |
|---|---|---|---|---|
| Discovery/mo | 5 | 50 | 200 | Unlimited |
| Companies/search | 10 | 25 | 25 | 50 |
| Contact reveals/mo | 0 | 100 | 500 | Unlimited |
| Saved leads | 10 | Unlimited | Unlimited | Unlimited |
| Members | 1 | 1 | 10 | Unlimited |
| Organizations | 1 | 1 | 1 | Unlimited |
| Batch | No | No | Yes | Yes |
| API | No | No | No | Yes |
| Auto-headhunt | No | Yes | Yes | Yes |
| AI Analysis | No | Yes | Yes | Yes |
| Shared workspace | No | No | Yes | Yes |
| Multi-org | No | No | No | Yes |
5.3 Credit Costs
| Operation | Credits | AI Cost (balanced) | Notes |
|---|---|---|---|
| Discovery search | 1 | ~$0.04 | 1 search = 15-25 companies |
| Contact reveal | 1 | ~$0.007-0.02 | 1 contact finding (headhunt) |
| Deep analysis (enrichment) | 2 | ~$0.003-0.02 | 1 company detailed analysis |
| Market intelligence | 3 | ~$0.06-0.10 | Multi-call pipeline |
| Batch (per company) | 0.5 | ~$0.00008 | Classify, minimal cost |
5.4 Credit Pack Prices
| Pack | Credits | Price | $/credit |
|---|---|---|---|
| Small | 50 | $12 | $0.24 |
| Medium | 200 | $45 | $0.225 |
| Large | 1,000 | $200 | $0.20 |
5.5 Credit Rollover
- Unused plan credits accumulate up to 2x the limit
- Example: Pro 100 reveals/mo → max 200 can accumulate, then new monthly credits stop adding
- Credit packs do not accumulate, valid for 12 months (expiry)
5.6 Annual Payment
- 20% discount (competitor average is 15-25%)
- Lemon Squeezy (MoR) charges the annual fee upfront in a single payment
- No card cancellation risk — payment already collected
- Refund policy: 30 days (LS standard)
6. Items to Finalize
This section must be resolved before beta.
- Token usage detail: Measure real average token count for each operation (from ai_job_runs table)
- Tier → model mapping needs optimization (cost vs quality testing)
- Sonar vs Sonar Pro quality comparison for headhunt
- GPT-4.1 Mini vs GPT-4o quality comparison for enrichment
- Credit rollover DB/code implementation (separate task)
- Annual plan needs to be defined in Lemon Squeezy (Phase C)
- export_ai_model_pricing table needs updating with current prices