Cheapest LLM APIs
Ranked by Price
Every major AI model sorted by cost. Updated daily from official pricing pages. 66 models tracked across 6 providers.
💡 Ultra-Budget Tier (under $0.50/1M input)
For high-volume, latency-tolerant workloads — classification, extraction, routing, chatbots.
| Rank | Model | Vendor | Input/1M | Output/1M | Best For |
|---|---|---|---|---|---|
| 1 | GPT-6 Luna | OpenAI | $0.1 | $0.5 | Ultra-cheap routing |
| 2 | Claude Haiku 5.5 | Anthropic | $0.1 | $0.5 | Ultra-cheap routing |
| 3 | Qwen3.6-35B-A3B | Qwen | $0.24 | $1.8 | Bulk processing |
| 4 | Gemini 3.1 Flash-Lite | $0.25 | $1.5 | Bulk processing | |
| 5 | Qwen3.5-122B-A10B | Qwen | $0.26 | $2.08 | Bulk processing |
| 6 | Gemini 3.5 Flash-Lite | $0.3 | $2.5 | Bulk processing | |
| 7 | DeepSeek V4.1 Flash | DeepSeek | $0.3 | $1.2 | Bulk processing |
| 8 | Qwen3.6-27B | Qwen | $0.3 | $3.2 | Bulk processing |
| 9 | Gemini 3.8 Flash TTS | $0.5 | $9.0 | Budget tasks |
⚙️ Mid-Tier ($0.50 - $5/1M input)
The workhorse range — best balance of capability and cost for most applications.
| Rank | Model | Vendor | Input/1M | Output/1M | Best For |
|---|---|---|---|---|---|
| 1 | Gemini 3.8 Flash | $0.75 | $3.75 | Coding & agents | |
| 2 | Gemini 3.7 Flash | $0.75 | $3.75 | Coding & agents | |
| 3 | Gemini 3.6 Flash | $0.75 | $3.75 | Coding & agents | |
| 4 | Gemini 3.8 Live | $0.75 | $4.5 | Coding & agents | |
| 5 | Claude Haiku 4.5 | Anthropic | $1 | $5 | Coding & agents |
| 6 | Grok Build 0.1 | xAI | $1.0 | $2.0 | Coding & agents |
| 7 | GPT-6.1 Terra | OpenAI | $1.0 | $5.0 | Coding & agents |
| 8 | Grok 4.3 | xAI | $1.25 | $2.5 | Coding & agents |
| 9 | GPT-4o Mini Transcribe | OpenAI | $1.25 | $5.0 | Coding & agents |
| 10 | DeepSeek V4 Pro | DeepSeek | $1.32 | $3.96 | Coding & agents |
| 11 | Gemini Omni Flash | $1.5 | $9.0 | Coding & agents | |
| 12 | GPT-5.3 Codex | OpenAI | $1.75 | $14.0 | Coding & agents |
| 13 | Claude Sonnet 5 | Anthropic | $2 | $10 | Coding & agents |
| 14 | Gemini 3.1 Pro | $2.0 | $12.0 | Coding & agents | |
| 15 | GPT-6 Sol | OpenAI | $2.0 | $10.0 | Coding & agents |
| 16 | Grok 4.7 | xAI | $2.0 | $6.0 | Coding & agents |
| 17 | Grok 4.6 | xAI | $2.0 | $6.0 | Coding & agents |
| 18 | Grok 4.5 | xAI | $2.0 | $6.0 | Coding & agents |
| 19 | Qwen3.8-2.4T-A95B | Qwen | $2.0 | $6.0 | Coding & agents |
| 20 | Claude Sonnet 5.5 | Anthropic | $2 | $10 | Coding & agents |
| 21 | GPT-6.1 Sol | OpenAI | $2.0 | $10.0 | Coding & agents |
| 22 | Gemini 4 Argon | $2.0 | $10.0 | Coding & agents | |
| 23 | GPT-4o Transcribe | OpenAI | $2.5 | $10.0 | General purpose |
| 24 | Claude Sonnet 4 | Anthropic | $3 | $15 | General purpose |
| 25 | Claude Sonnet 4.6 | Anthropic | $3 | $15 | General purpose |
| 26 | Claude Sonnet 4.5 | Anthropic | $3 | $15 | General purpose |
| 27 | Claude Opus 5.5 | Anthropic | $4 | $20 | General purpose |
| 28 | GPT-5.6 Sol | OpenAI | $4.0 | $20.0 | General purpose |
| 29 | Claude Opus 5 | Anthropic | $5 | $25 | Complex tasks |
| 30 | Claude Opus 4.5 | Anthropic | $5 | $25 | Complex tasks |
| 31 | Claude Opus 4.8 | Anthropic | $5 | $25 | Complex tasks |
| 32 | Claude Opus 4.7 | Anthropic | $5 | $25 | Complex tasks |
| 33 | Claude Opus 4.6 | Anthropic | $5 | $25 | Complex tasks |
| 34 | GPT Rosalind Research | OpenAI | $5.0 | $25.0 | Complex tasks |
| 35 | GPT Rosalind Discovery | OpenAI | $5.0 | $25.0 | Complex tasks |
| 36 | Chat Latest | OpenAI | $5.0 | $30.0 | Complex tasks |
🚀 Premium Tier (over $5/1M input)
For maximum capability — research, complex reasoning, specialized workloads.
| Rank | Model | Vendor | Input/1M | Output/1M | Best For |
|---|---|---|---|---|---|
| 1 | GPT Image 2.5 | OpenAI | $8.0 | $30.0 | Flagship tasks |
| 2 | Claude Fable 5.1 | Anthropic | $10 | $50 | Flagship tasks |
| 3 | Claude Mythos 5.1 | Anthropic | $10 | $50 | Flagship tasks |
| 4 | GPT-6 Astra | OpenAI | $10.0 | $50.0 | Flagship tasks |
| 5 | GPT Realtime 2.1 Mini | OpenAI | $10.0 | $20.0 | Flagship tasks |
| 6 | Claude Fable 5 | Anthropic | $10 | $50 | Flagship tasks |
| 7 | Claude Mythos 5 | Anthropic | $10 | $50 | Flagship tasks |
| 8 | GPT-6 Astra Law | OpenAI | $10.0 | $50.0 | Flagship tasks |
| 9 | GPT-5.6 Cyber | OpenAI | $12.5 | $75.0 | Flagship tasks |
| 10 | Claude Opus 4 | Anthropic | $15 | $75 | Ultra-premium |
| 11 | Claude Opus 4.1 | Anthropic | $15 | $75 | Ultra-premium |
| 12 | GPT-6 Rosalind | OpenAI | $15.0 | $75.0 | Ultra-premium |
| 13 | GPT-6 Cyber | OpenAI | $20.0 | $100.0 | Ultra-premium |
| 14 | GPT Realtime 2.1 | OpenAI | $32.0 | $64.0 | Ultra-premium |
📊 Quick Comparison: Best Value at Each Level
Most Bang for Buck
| Category | Winner | Why |
|---|---|---|
| Absolute cheapest | Claude Haiku 5.5 / GPT-6 Luna | Both $0.10/$0.50 — tied for cheapest |
| Best budget value | GPT-6.1 Terra | $1/$5 — half of Sol, 10x Luna, full GPT-6 quality |
| Best mid-tier value | DeepSeek V4.1 Flash | $0.30/$1.20 + off-peak 50% = $0.15/$0.60 |
| Best flagship value | GPT-6 Astra | $10/$50 — same as Argon's NEW (doubled) rate |
| Biggest recent price cut | Claude Opus 4.5-4.8 | $15/$75 → $5/$25 (-67%) |
Frequently Asked Questions
What is the cheapest LLM API in 2026?
Claude Haiku 5.5 and GPT-6 Luna are tied at $0.10/$0.50 per million tokens. DeepSeek V4.1 Flash is close at $0.30/$1.20 (or $0.15/$0.60 off-peak).
Which LLM API has the best value?
For most workloads, GPT-6.1 Terra ($1/$5) or DeepSeek V4.1 Flash ($0.30/$1.20) offer the best capability-per-dollar. For premium tasks, GPT-6 Astra ($10/$50) or Claude Opus 5.5 ($4/$20) after Anthropic's 67% price cut.
Are AI API prices going up or down?
Both. Google is raising prices (Gemini doubles on Dec 1 and Jan 1) while Anthropic is cutting (Opus 4.5-4.8 dropped 67%). See our price increase tracker for details.
How can I reduce my LLM API costs?
1) Use Batch API (50% off, available from OpenAI, Anthropic, Google). 2) Enable prompt caching (up to 90% cheaper on repeated context). 3) Use off-peak hours (DeepSeek: 50% off). 4) Downgrade to a cheaper model for simple tasks. 5) See our upcoming increases to plan ahead.
Sorted by standard input price per 1M tokens. Prices tracked daily from official pages. Last updated 2026-11-06. Some models have off-peak, batch, or promotional rates that may be cheaper — see individual model pages for details.