LLM API Pricing Comparison
Compare pricing across 424+ LLM APIs from OpenAI, Anthropic, Google, DeepSeek, Mistral, xAI, and more. Sorted by quality, price, or value score.
Pricing TLDR
- • Live pricing from $0.01/M input up to $180/M output
- • Quality scores from 0-100, based on independent benchmarks (Theozard)
- • Value score = quality per dollar of output cost, so you can spot the best value fast
Monthly LLM API Cost Comparison
Calculate by
Input Tokens
Output Tokens
API Calls / Month
Quick Examples:
Sort:
(anthropic/claude-opus-5.5:batch)
Context
Quality
Per 1M Tokens
In: $2.00
Out: $10.00
Value
Monthly Cost
(anthropic/claude-opus-5.5)
Context
Quality
Per 1M Tokens
In: $4.00
Out: $20.00
Value
Monthly Cost
(openai/gpt-6-astra:batch)
Context
Quality
Per 1M Tokens
In: $5.00
Out: $25.00
Value
Monthly Cost
(openai/gpt-6-astra)
Context
Quality
Per 1M Tokens
In: $10.00
Out: $50.00
Value
Monthly Cost
(openai/gpt-6.1-sol:batch)
Context
Quality
Per 1M Tokens
In: $1.00
Out: $5.00
Value
Monthly Cost
(anthropic/claude-sonnet-5.5:batch)
Context
Quality
Per 1M Tokens
In: $1.00
Out: $5.00
Value
Monthly Cost
(openai/gpt-6.1-sol)
Context
Quality
Per 1M Tokens
In: $2.00
Out: $10.00
Value
Monthly Cost
(anthropic/claude-sonnet-5.5)
Context
Quality
Per 1M Tokens
In: $2.00
Out: $10.00
Value
Monthly Cost
(openai/gpt-6.1-terra:batch)
Context
Quality
Per 1M Tokens
In: $0.50
Out: $2.50
Value
Monthly Cost
(openai/gpt-6.1-terra)
Context
Quality
Per 1M Tokens
In: $1.00
Out: $5.00
Value
Monthly Cost
(anthropic/claude-opus-5:batch)
Context
Quality
Per 1M Tokens
In: $2.50
Out: $12.50
Value
Monthly Cost
(anthropic/claude-opus-5)
Context
Quality
Per 1M Tokens
In: $5.00
Out: $25.00
Value
Monthly Cost
(google/gemini-4-argon:batch)
Context
Quality
Per 1M Tokens
In: $1.00
Out: $5.00
Value
Monthly Cost
(google/gemini-4-argon)
Context
Quality
Per 1M Tokens
In: $2.00
Out: $10.00
Value
Monthly Cost
(anthropic/claude-fable-5.1:batch)
Context
Quality
Per 1M Tokens
In: $5.00
Out: $25.00
Value
Monthly Cost
(anthropic/claude-fable-5:batch)
Context
Quality
Per 1M Tokens
In: $5.00
Out: $25.00
Value
Monthly Cost
(anthropic/claude-fable-5.1)
Context
Quality
Per 1M Tokens
In: $10.00
Out: $50.00
Value
Monthly Cost
(anthropic/claude-fable-5)
Context
Quality
Per 1M Tokens
In: $10.00
Out: $50.00
Value
Monthly Cost
(moonshotai/kimi-k3.1-pro)
Context
Quality
Per 1M Tokens
In: $2.00
Out: $10.00
Value
Monthly Cost
(openai/gpt-5.6-sol:batch)
Context
Quality
Per 1M Tokens
In: $2.00
Out: $10.00
Value
Monthly Cost
(openai/gpt-5.6-sol)
Context
Quality
Per 1M Tokens
In: $4.00
Out: $20.00
Value
Monthly Cost
(meta/muse-spark-1.3)
Context
Quality
Per 1M Tokens
In: $1.25
Out: $4.25
Value
Monthly Cost
(x-ai/grok-4.6:batch)
Context
Quality
Per 1M Tokens
In: $1.60
Out: $4.80
Value
Monthly Cost
(openai/gpt-6-sol:batch)
Context
Quality
Per 1M Tokens
In: $1.00
Out: $5.00
Value
Monthly Cost
(x-ai/grok-4.6)
Context
Quality
Per 1M Tokens
In: $2.00
Out: $6.00
Value
Monthly Cost
(openai/gpt-6-sol)
Context
Quality
Per 1M Tokens
In: $2.00
Out: $10.00
Value
Monthly Cost
(moonshotai/kimi-k3:batch)
Context
Quality
Per 1M Tokens
In: $2.28
Out: $11.40
Value
Monthly Cost
(moonshotai/kimi-k3)
Context
Quality
Per 1M Tokens
In: $0.19
Out: $16.00
Value
Monthly Cost
(meta/muse-wonder)
Context
Quality
Per 1M Tokens
In: $4.00
Out: $20.00
Value
Monthly Cost
(anthropic/claude-opus-4.8:batch)
Context
Quality
Per 1M Tokens
In: $2.50
Out: $12.50
Value
Monthly Cost
Spending across 315+ OpenRouter models?
Monitor your OpenRouter credits and usage in real-time.
Free 7-day trial. No sign-up, no credit card.
Best Value LLM APIs by Quality Per Dollar
Value score = quality points per $1 of output cost (per 1M tokens). Higher is better. Only models above a usable quality bar are ranked, so cheap low-quality models don't top the list.
#
Model
inclusionAI: Ling 3.1 Flash
Provider
Quality
Output / 1M
Value Score
#
Model
DeepSeek: DeepSeek V4 Flash 0731 (batch)
Provider
Quality
Output / 1M
Value Score
#
Model
DeepSeek: DeepSeek V4.1 Flash 1017 (batch)
Provider
Quality
Output / 1M
Value Score
#
Model
Anthropic: Claude Haiku 5.5 (batch)
Provider
Quality
Output / 1M
Value Score
#
Model
DeepSeek: DeepSeek V4.1 Flash (batch)
Provider
Quality
Output / 1M
Value Score
#
Model
DeepSeek: DeepSeek V4 Flash 0731
Provider
Quality
Output / 1M
Value Score
#
Model
Anthropic: Claude Haiku 5.5
Provider
Quality
Output / 1M
Value Score
#
Model
DeepSeek: DeepSeek V4.1 Flash
Provider
Quality
Output / 1M
Value Score
#
Model
DeepSeek: DeepSeek V4.1 Flash 1017
Provider
Quality
Output / 1M
Value Score
#
Model
OpenAI: GPT-5.6 Luna (batch)
Provider
Quality
Output / 1M
Value Score
About LLM API Pricing
What is LLM API Pricing?
LLM APIs let you integrate large language models into your applications via HTTP requests. Every major AI provider (OpenAI, Anthropic, Google, DeepSeek, Mistral, xAI) offers API access to their models with per-token pricing. You pay separately for input tokens (your prompts) and output tokens (model responses), quoted per million tokens.
- Input vs Output Token Pricing: Input tokens (prompts, context) are cheaper because they only need to be processed once. Output tokens (completions) cost 2-5x more because each token requires a full forward pass through the model. Optimizing prompt length has the biggest impact on cost.
- Quality-Price Tradeoff: More expensive models generally deliver higher quality. Our quality scores (0-100) let you compare directly: the top models sit near 100 while budget models score lower but cost a fraction as much. The right choice depends on how much quality your task actually needs.
- Context Window Costs: Larger context windows let you send more data per request but increase token costs. A 200K context model processing long documents costs proportionally more in input tokens than a short chatbot interaction. Choose context size based on your actual needs.
When to Use LLM API Pricing
Different use cases call for different models. Match your quality requirements to your budget using the value score to find the optimal model.
Ideal for
- Chatbots and conversational AI, where mid-tier models offer the best quality-to-cost balance
- Code generation, where code-specialized models optimize for the task
- Bulk content processing, where budget models handle volume at minimal cost
- Complex reasoning, where premium flagship models justify their cost on hard problems
- Prototyping, where free-tier models let you build at no cost
Not ideal for
- Real-time applications needing sub-100ms latency (consider edge-deployed models)
- Tasks that don't need language understanding (use traditional algorithms instead)
- Processing sensitive data with compliance requirements (check each provider's data policies)
LLM API Monthly Cost Estimates
Hobby / Prototyping
$0-10/mo
• Free tier models
• < 1K requests/day
• Testing & development
Startup / MVP
$50-300/mo
• Mid-tier models
• 5-20K requests/day
• Single product
Growth
$300-2,000/mo
• Premium & budget mix
• 20-100K requests/day
• Multiple use cases
Enterprise
$2,000+/mo
• Premium models
• 100K+ requests/day
• Model fallback chains
5 LLM API Cost Optimization Tips
Use a Model Cascade
Route easy queries to cheap models and only escalate to premium ones when needed. A classifier model can decide the routing. This typically saves 60-80% versus using premium models for everything.
Optimize Prompt Length
Input tokens cost money. Strip unnecessary context, use concise system prompts, and avoid sending full documents when a summary suffices. A 50% reduction in prompt length = 50% savings on input costs.
Cache Frequent Requests
If you make similar API calls repeatedly, cache responses. Many providers also offer prompt caching features that reduce costs for repeated system prompts. Anthropic's prompt caching can save up to 90% on cached tokens.
Value Score Beats Lowest Price
The cheapest model isn't always the best value. A rock-bottom price with a low quality score can deliver less than a mid-price model with a much higher score. Use the value-score column to find the sweet spot for your needs.
Monitor Per-Model Spending
Track costs per model and per use case with CostGoat. Identify which models consume the most budget, find opportunities to downgrade specific workflows, and catch cost spikes early before they become expensive surprises.
Start Tracking Your LLM API Spending
Monitor spending across OpenAI, Anthropic, Google, OpenRouter, and other LLM providers, all from one menubar app.
Free 7-day trial. No sign-up, no credit card.
LLM API Pricing FAQ
Common questions about LLM API costs, pricing models, and how to save money
LLM APIs charge per token, with separate rates for input tokens (your prompts) and output tokens (model responses), quoted per million tokens. Output tokens cost more than input because each one requires a full pass through the model. Most providers offer pay-as-you-go with no minimum commitments.
Meta: Llama 3.1 8B Instruct at $0.02/$0.04 per 1M tokens (input/output) is among the cheapest options right now. If you care about quality per dollar rather than raw price, check the value-score column, where inclusionAI: Ling 3.1 Flash, DeepSeek: DeepSeek V4 Flash 0731 (batch), DeepSeek: DeepSeek V4.1 Flash 1017 (batch) tend to lead. Many providers also offer free tiers with rate limits on OpenRouter.
OpenAI and Anthropic price comparably tier for tier. Their flagship models command premium per-token rates, their mid-tier models land in the low single digits per million input tokens, and both offer budget models at a fraction of that. Exact prices shift often, so use the comparison table above for current rates.
Yes, usually by a wide margin. DeepSeek's open models price their output tokens far below the OpenAI flagships, often by one to two orders of magnitude. The trade-off is lower benchmark quality, though the gap has narrowed. For chatbots and content generation the value is strong, so use the value-score column to compare directly.
The value score in our comparison (quality points per dollar of output cost) surfaces the models that deliver the most quality per dollar. Right now inclusionAI: Ling 3.1 Flash, DeepSeek: DeepSeek V4 Flash 0731 (batch), DeepSeek: DeepSeek V4.1 Flash 1017 (batch) tend to top the value rankings. For premium quality at mid-range prices, sort the table by value score and scan the higher-quality rows.
