Start free
Use zero-price rows only for prototypes, demos, and fallback tests. Confirm rate limits before public launch.
Free Tool · Pricing Monitor · Updated CSV/JSON
Cheapest LLM API Leaderboard
A practical AI API pricing monitor for builders. Compare tracked LLM models by input/output token price, estimated monthly cost, free-credit signals, signup friction, OpenAI compatibility, and China-friendly access.
14priced model rows
$1.32cheapest paid sample month
11no-card / low-friction rows
11OpenAI-compatible rows
Methodology snapshot
This benchmark uses a fixed workload of 10M input tokens + 2M output tokens per month. Rows are sorted by estimated monthly API cost, then enriched with signup friction, free-credit, regional access, and OpenAI-compatible signals.
Use it for shortlisting, not final billing. Provider prices, free credits, exchange rates, and limits change often; always verify official pricing before production.
Interactive monitor
Re-rank by your monthly token workload
Visible rows14
Cheapest visible—
Estimated month—
The table below recalculates locally in your browser. Use the CSV/JSON export when citing the default benchmark; use this panel to sanity-check your own workload.
Citeable table
AI API pricing monitor
| Rank | Model | Input / 1M | Output / 1M | Sample month | Access | Signup / credits |
|---|---|---|---|---|---|---|
| #1 | OpenRouter Free ModelsOpenRouter · profile · 2026-05-16varies context · free | $0 | $0 | $0customizable workload | Needs testingOpenAI-compatible | Some models are freeNo-card / low-friction signal |
| #2 | GLM-4 FlashZhipu GLM · profile · 2026-05-16128k context · free | $0 | $0 | $0customizable workload | China-friendlyOpenAI-compatible | Check BigModel dashboardNo-card / low-friction signal |
| #3 | Doubao LiteDoubao · profile · 2026-05-16varies context · budget | $0.11 | $0.11 | $1.32customizable workload | China-friendlyOpenAI-compatible | Check Volcano Ark dashboardNo-card / low-friction signal |
| #4 | Hunyuan LiteTencent Hunyuan · profile · 2026-05-16varies context · budget | $0.14 | $0.14 | $1.68customizable workload | China-friendlyOpenAI-compatible | Check Tencent Cloud dashboardNo-card / low-friction signal |
| #5 | SiliconFlow DeepSeek/Qwen CompatibleSiliconFlow · profile · 2026-05-16varies context · budget | $0.14 | $0.28 | $1.96customizable workload | China-friendlyOpenAI-compatible | Check SiliconFlow dashboardNo-card / low-friction signal |
| #6 | DeepSeek ChatDeepSeek · profile · 2026-05-1664k context · budget | $0.27 | $1.10 | $4.90customizable workload | China-friendlyOpenAI-compatible | Check dashboardNo-card / low-friction signal |
| #7 | 上海电信 25 万额度点套餐Shanghai Telecom Token Package · profile · 2026-05-16multi-model context · telco-package | $0.57 | $0.57 | $6.84customizable workload | China-friendlyPartial compatibility | Reported ¥1 for about 250k quota points, mobile-bill paymentNo-card / low-friction signal |
| #8 | GPT-4.1 miniOpenAI · profile · 2026-05-161M context · balanced | $0.40 | $1.60 | $7.20customizable workload | Limited / relay likelyOpenAI-compatible | $5Card likely required |
| #9 | Llama 3.3 70B on GroqGroq · profile · 2026-05-16128k context · fast | $0.59 | $0.79 | $7.48customizable workload | Needs testingOpenAI-compatible | Free tier, rate limits varyNo-card / low-friction signal |
| #10 | Gemini 2.5 FlashGoogle Gemini · profile · 2026-05-161M context · balanced | $0.30 | $2.50 | $8.00customizable workload | Limited / relay likelyPartial compatibility | Free tier, model and region limits varyNo-card / low-friction signal |
| #11 | DeepSeek ReasonerDeepSeek · profile · 2026-05-1664k context · reasoning | $0.55 | $2.19 | $9.88customizable workload | China-friendlyOpenAI-compatible | Check dashboardNo-card / low-friction signal |
| #12 | Moonshot v1 8KKimi / Moonshot · profile · 2026-05-168k context · balanced | $1.68 | $1.68 | $20customizable workload | China-friendlyOpenAI-compatible | Check dashboardNo-card / low-friction signal |
| #13 | GPT-4.1OpenAI · profile · 2026-05-161M context · premium | $2.00 | $8.00 | $36customizable workload | Limited / relay likelyOpenAI-compatible | $5Card likely required |
| #14 | Claude Sonnet 4Anthropic Claude · profile · 2026-05-16200k context · premium | $3.00 | $15 | $60customizable workload | Limited / relay likelyProvider SDK | $0Card likely required |
Reusable citation block
Suggested citation: “Yangmao AI API Pricing Monitor compares 14 tracked LLM API model rows by input/output token price and a 10M input + 2M output monthly workload. Source: yangmao.ai, dataset updated 2026-06-24, source snapshot generated 2026-06-24 from public provider/pricing records.”
Canonical: https://yangmao.ai/en/tools/cheapest-llm-api-leaderboard/. Default benchmark: 10,000,000 input tokens + 2,000,000 output tokens/month.
For citations and backlinks
Dataset dictionary and reuse rules
Field definitions
- rank
- Ascending rank by sample_monthly_cost_usd for the default benchmark workload.
- input_price_per_1m
- Published or tracked input-token price per 1 million tokens, normalized to USD where available.
- output_price_per_1m
- Published or tracked output-token price per 1 million tokens, normalized to USD where available.
- sample_monthly_cost_usd
- Estimated API spend for 10M input tokens plus 2M output tokens per month.
- china_access
- china_friendly, relay_likely, or needs_testing signal for access from China-oriented teams.
- requires_card
- Boolean signup-friction signal from tracked provider/free-credit metadata.
- openai_compatible
- yes, partial, or no based on tracked provider API compatibility metadata.
- utm_cta_url
- Yangmao-owned CTA URL for routing/evaluation attribution from reused dataset rows.
Reuse checklist
- Link to the canonical leaderboard when quoting ranks or sample monthly cost.
- Include generated_at and the default workload when republishing rows.
- Treat zero-price and no-card rows as shortlist signals, not guaranteed production terms.
Every model row includes a provider page, calculator URL, citation label, last-checked field, and OpenLLMAPI UTM CTA for attributed follow-up traffic.
Decision shortcuts
Low-friction migration
Prioritize no-card plus OpenAI-compatible rows when you need a fast drop-in test.
China-friendly shortlist
Use China-friendly rows when signup, payment, and endpoint reachability matter more than global brand coverage.