Monthly index
India AI API Price Index, October 2026
What it costs to call an AI model from India this month, in rupees per million tokens, across 419 paid chat models from 41 authors. Prices are the provider list prices Rapid Router bills at, converted at ₹98.86 per dollar. Free to cite with a link (CC BY 4.0).
Blended = 75% input price + 25% output price per million tokens, the mix of a typical chat or RAG workload.
The 25 cheapest paid models this month
| # | Model | Input / 1M | Output / 1M | Blended / 1M |
|---|---|---|---|---|
| 1 | Llama 3.2 1B InstructMeta | ₹0.494 | ₹0.989 | ₹0.618 |
| 2 | Llama 3.2 3B InstructMeta | ₹1.98 | ₹1.98 | ₹1.98 |
| 3 | Qwen3.5 0.8BQwen | ₹0.989 | ₹4.94 | ₹1.98 |
| 4 | Meta Llama 3.1 8B Instruct TurboMeta | ₹1.98 | ₹3.95 | ₹2.47 |
| 5 | Meta Llama 3.1 8B Instruct Turbo (free)Meta | ₹1.98 | ₹3.95 | ₹2.47 |
| 6 | Llama-3.1-8B-InstructMeta | ₹1.98 | ₹4.94 | ₹2.72 |
| 7 | Llama-3.1-8B-InstructNovita | ₹1.98 | ₹4.94 | ₹2.72 |
| 8 | Mistral NemoMistral | ₹2.87 | ₹2.97 | ₹2.89 |
| 9 | Mistral Nemo (free)Mistral | ₹2.87 | ₹2.97 | ₹2.89 |
| 10 | DeepSeek OCR 2Novita | ₹2.97 | ₹2.97 | ₹2.97 |
| 11 | Qwen2.5 Coder 7BQwen | ₹2.47 | ₹4.94 | ₹3.09 |
| 12 | Meta Llama 3 8B InstructMeta | ₹2.97 | ₹3.95 | ₹3.21 |
| 13 | mistral-7b-instruct-v0.3Mistral | ₹2.77 | ₹5.34 | ₹3.41 |
| 14 | Gemma 2 9BGoogle | ₹2.97 | ₹5.93 | ₹3.71 |
| 15 | L3 8B Lunaris V1Sao10k | ₹2.97 | ₹5.93 | ₹3.71 |
| 16 | Gemma 4 E4B ITGoogle | ₹1.98 | ₹9.89 | ₹3.95 |
| 17 | qwen3.5-2bQwen | ₹1.98 | ₹9.89 | ₹3.95 |
| 18 | L3 8B Lunaris V1 TurboSao10k | ₹3.95 | ₹4.94 | ₹4.20 |
| 19 | DeepSeek OCRDeepseek Ai | ₹2.97 | ₹9.89 | ₹4.70 |
| 20 | Sao10K Stheno 8bNovita | ₹4.94 | ₹4.94 | ₹4.94 |
| 21 | Llama-Guard-3-8B (Deep Infra)Meta | ₹5.44 | ₹5.44 | ₹5.44 |
| 22 | WizardLM 2 7BMicrosoft | ₹5.44 | ₹5.44 | ₹5.44 |
| 23 | Mistral 7B Instruct V0.1Mistral | ₹5.44 | ₹5.44 | ₹5.44 |
| 24 | mistral-7b-instruct-v0.2Mistral | ₹5.44 | ₹5.44 | ₹5.44 |
| 25 | OpenChat 3.6 8BOpenchat | ₹5.44 | ₹5.44 | ₹5.44 |
Flagships, priced in rupees
| Model | Input / 1M | Output / 1M | 1M-token chat (75/25) |
|---|---|---|---|
| Qwen3.8 MaxNovita | ₹198 | ₹593 | ₹297 |
| Qwen3.8 2.4T A95BQwen | ₹198 | ₹593 | ₹297 |
| Grok 4.5xAI | ₹198 | ₹593 | ₹297 |
| Grok 4.6xAI | ₹198 | ₹593 | ₹297 |
| Grok 4.7xAI | ₹198 | ₹593 | ₹297 |
| GPT-3.5 Turbo 16kOpenAI | ₹297 | ₹395 | ₹321 |
| GPT 4.1OpenAI | ₹198 | ₹791 | ₹346 |
| GPT 4.1OpenAI | ₹198 | ₹791 | ₹346 |
| o3OpenAI | ₹198 | ₹791 | ₹346 |
| o3OpenAI | ₹198 | ₹791 | ₹346 |
| Qwen-MT PlusQwen | ₹243 | ₹729 | ₹365 |
| Claude Sonnet 5Anthropic | ₹198 | ₹989 | ₹395 |
| Claude Sonnet 5.5Anthropic | ₹198 | ₹989 | ₹395 |
| CommunityNovita | ₹395 | ₹395 | ₹395 |
| GPT 6 SolOpenAI | ₹198 | ₹989 | ₹395 |
| GPT 6.1 SolOpenAI | ₹198 | ₹989 | ₹395 |
| Nemotron 4 340B InstructNVIDIA | ₹415 | ₹415 | ₹415 |
| GPT-4oOpenAI | ₹247 | ₹989 | ₹433 |
| GPT-4oOpenAI | ₹247 | ₹989 | ₹433 |
| GPT-4oOpenAI | ₹247 | ₹989 | ₹433 |
How to read this
Every number is a provider list price, the same figure a direct account with that provider would charge in dollars, converted to rupees at the rate Rapid Router bills at today. Paying from India through a provider directly adds a card forex markup and 18% GST under reverse charge on top of the dollar figure; through Rapid Router the GST is on the invoice and the platform fee is charged once when you buy credits, not per request.
Tiers are by input price: flagship at $2 or more per million, mid-range from $0.40, small below that. Reasoning and vision follow each model's published capabilities. Open-weight covers models whose weights are publicly released, whoever serves them. Free-tier models are excluded; they are listed on the free tier page.
To cite: "India AI API Price Index, October 2026, Rapid Router (rapidrouter.ai/price-index)". The underlying data is the catalog snapshot, refreshed on every release. For a monthly bill for your own volume, use the cost calculator.