Data
One OCR'd invoice in; structured JSON out. Priced per document and per thousand.
Pricing verified:
Between $0.0000 and $0.013 per run, depending on the model. That is a 409× spread for the same job.
A single-page OCR'd invoice with line items ≈ 1,200 input tokens. A structured JSON record of the fields ≈ 300 output tokens.
Cost = (input tokens ÷ 1M × input price) + (output tokens ÷ 1M × output price). Prices come from our daily-verified model data. Batch and cached-input discounts are not applied — they only ever make these numbers smaller.
| Model | Provider | Per run | At 1,000 invoices a month | Context |
|---|---|---|---|---|
| Mistral: Mistral Nemo | Mistral | $0.0000 | $0.033 | 131k tokens |
| Google: Gemma 2 9B | $0.0001 | $0.063 | 8k tokens | |
| Meta: Llama 3.1 8B Instruct | Meta | $0.0001 | $0.084 | 16k tokens |
| OpenAI: GPT-5 Nano | OpenAI | $0.0001 | $0.090 | 400k tokens |
| Meta: Llama 3.2 1B Instruct | Meta | $0.0001 | $0.092 | 60k tokens |
| OpenAI: GPT-4.1 Nano | OpenAI | $0.0001 | $0.120 | 1.0M tokens |
| Mistral: Ministral 3 3B 2512 | Mistral | $0.0001 | $0.150 | 131k tokens |
| Mistral: Mistral Small 3.2 24B | Mistral | $0.0001 | $0.150 | 128k tokens |
| GPT-4o | OpenAI | $0.0002 | $0.180 | 128k tokens |
| Google: Gemini 2.0 Flash Lite | $0.0002 | $0.180 | 1.0M tokens | |
| OpenAI: gpt-oss-safeguard-20b | OpenAI | $0.0002 | $0.180 | 131k tokens |
| Mistral: Mistral 7B Instruct v0.1 | Mistral | $0.0002 | $0.189 | 3k tokens |
| Llama 4 Scout | Meta | $0.0002 | $0.210 | 512k tokens |
| Mistral: Devstral Small 1.1 | Mistral | $0.0002 | $0.210 | 131k tokens |
| Mistral: Mistral Small Creative | Mistral | $0.0002 | $0.210 | 33k tokens |
| Mistral: Voxtral Small 24B 2507 | Mistral | $0.0002 | $0.210 | 32k tokens |
| Meta: Llama 3 8B Instruct | Meta | $0.0002 | $0.210 | 8k tokens |
| Mistral: Ministral 3 8B 2512 | Mistral | $0.0002 | $0.225 | 262k tokens |
| Google: Gemini 2.0 Flash | $0.0002 | $0.240 | 1.0M tokens | |
| Google: Gemini 2.5 Flash Lite | $0.0002 | $0.240 | 1.0M tokens | |
| Google: Gemini 2.5 Flash Lite Preview 09-2025 | $0.0002 | $0.240 | 1.0M tokens | |
| DeepSeek V4-Flash | DeepSeek | $0.0003 | $0.252 | 1M tokens |
| Meta: Llama Guard 4 12B | Meta | $0.0003 | $0.270 | 164k tokens |
| Gemma 4 26B A4B | $0.0003 | $0.276 | 262k tokens | |
| Gemma 4 31B | $0.0003 | $0.288 | 262k tokens | |
| Mistral: Ministral 3 14B 2512 | Mistral | $0.0003 | $0.300 | 262k tokens |
| GPT-5.6 Luna | OpenAI | $0.0003 | $0.300 | 1.1M tokens |
| GPT-4o Mini | OpenAI | $0.0004 | $0.360 | 128k tokens |
| Mistral: Mistral Small 4 | Mistral | $0.0004 | $0.360 | 262k tokens |
| Mistral: Saba | Mistral | $0.0004 | $0.420 | 33k tokens |
| Llama 4 Maverick | Meta | $0.0005 | $0.480 | 256k tokens |
| OpenAI: GPT-4.1 Mini | OpenAI | $0.0005 | $0.480 | 1.0M tokens |
| xAI: Grok 3 Mini | xAI | $0.0005 | $0.510 | 131k tokens |
| xAI: Grok 3 Mini Beta | xAI | $0.0005 | $0.510 | 131k tokens |
| Meta: Llama 3.2 11B Vision Instruct | Meta | $0.0005 | $0.517 | 131k tokens |
| OpenAI: GPT-3.5 Turbo | OpenAI | $0.0005 | $0.525 | 16k tokens |
| Google: Gemini 2.5 Flash | $0.0006 | $0.555 | 1.0M tokens | |
| Llama Guard 3 8B | Meta | $0.0006 | $0.585 | 131k tokens |
| Mistral Small 3.1 | Mistral | $0.0006 | $0.588 | 128k tokens |
| Mistral: Mistral Small 3 | Mistral | $0.0006 | $0.588 | 33k tokens |
| Meta: Llama 3.1 70B Instruct | Meta | $0.0006 | $0.600 | 131k tokens |
| Mistral: Codestral 2508 | Mistral | $0.0006 | $0.630 | 256k tokens |
| DeepSeek V3 | DeepSeek | $0.0007 | $0.654 | 128k tokens |
| Anthropic: Claude 3 Haiku | Anthropic | $0.0007 | $0.675 | 200k tokens |
| xAI: Grok Code Fast 1 | xAI | $0.0007 | $0.690 | 256k tokens |
| Google: Gemini 3 Flash Preview | $0.0008 | $0.750 | 1.0M tokens | |
| DeepSeek V4-Pro | DeepSeek | $0.0008 | $0.783 | 1M tokens |
| Mistral: Mixtral 8x7B Instruct | Mistral | $0.0008 | $0.810 | 33k tokens |
| Meta: Llama 3 70B Instruct | Meta | $0.0008 | $0.834 | 8k tokens |
| OpenAI: GPT-5 Mini | OpenAI | $0.0009 | $0.900 | 400k tokens |
| OpenAI: GPT-5.1-Codex-Mini | OpenAI | $0.0009 | $0.900 | 400k tokens |
| Google: Gemma 2 27B | $0.0010 | $0.975 | 8k tokens | |
| Gemini 3.6 Flash | $0.0010 | $1.01 | 1.0M tokens | |
| Mistral: Mistral Large 3 2512 | Mistral | $0.0010 | $1.05 | 262k tokens |
| Mistral: Devstral 2 2512 | Mistral | $0.0011 | $1.08 | 262k tokens |
| Mistral: Devstral Medium | Mistral | $0.0011 | $1.08 | 131k tokens |
| Mistral: Mistral Medium 3 | Mistral | $0.0011 | $1.08 | 131k tokens |
| Mistral: Mistral Medium 3.1 | Mistral | $0.0011 | $1.08 | 131k tokens |
| Gemini 3.5 Flash-Lite | $0.0011 | $1.11 | 1.0M tokens | |
| Google: Nano Banana (Gemini 2.5 Flash Image) | $0.0011 | $1.11 | 33k tokens | |
| DeepSeek R1 | DeepSeek | $0.0013 | $1.32 | 128k tokens |
| OpenAI: o3 Mini | OpenAI | $0.0013 | $1.32 | 200k tokens |
| OpenAI: o3 Mini High | OpenAI | $0.0013 | $1.32 | 200k tokens |
| OpenAI: o4 Mini | OpenAI | $0.0013 | $1.32 | 200k tokens |
| OpenAI: o4 Mini High | OpenAI | $0.0013 | $1.32 | 200k tokens |
| Anthropic: Claude Haiku 4.5 | Anthropic | $0.0013 | $1.35 | 200k tokens |
| OpenAI: GPT Audio Mini | OpenAI | $0.0014 | $1.44 | 128k tokens |
| Gemini 3.1 Flash | $0.0015 | $1.50 | 1M tokens | |
| OpenAI: GPT-3.5 Turbo (older v0613) | OpenAI | $0.0018 | $1.80 | 4k tokens |
| Codestral 25.01 | Mistral | $0.0019 | $1.89 | 256k tokens |
| Anthropic: Claude 3.5 Haiku | Anthropic | $0.0022 | $2.16 | 200k tokens |
| Claude 4 Haiku | Anthropic | $0.0022 | $2.16 | 200k tokens |
| Gemini 3.5 Flash | $0.0022 | $2.25 | 1.0M tokens | |
| Google: Gemini 2.5 Pro | $0.0022 | $2.25 | 1.0M tokens | |
| OpenAI: GPT-5 Codex | OpenAI | $0.0022 | $2.25 | 400k tokens |
| Kimi K2.7 Code | Moonshot | $0.0023 | $2.34 | 256k tokens |
| OpenAI: GPT-3.5 Turbo Instruct | OpenAI | $0.0024 | $2.40 | 4k tokens |
| OpenAI: GPT-4.1 | OpenAI | $0.0024 | $2.40 | 1.0M tokens |
| OpenAI: o3 | OpenAI | $0.0024 | $2.40 | 200k tokens |
| Claude Sonnet 5 | Anthropic | $0.0027 | $2.70 | 1M tokens |
| GPT-5.6 Sol | OpenAI | $0.0027 | $2.70 | 1.1M tokens |
| Muse Spark | Meta | $0.0028 | $2.77 | 1.0M tokens |
| GPT-5.2 Mini | OpenAI | $0.0029 | $2.88 | 128k tokens |
| GLM-5.2 | Z.ai | $0.0030 | $3.00 | 1M tokens |
| GPT-5.6 Terra | OpenAI | $0.0030 | $3.00 | 1.1M tokens |
| OpenAI: GPT-5 | OpenAI | $0.0030 | $3.00 | 400k tokens |
| OpenAI: GPT-5 Image Mini | OpenAI | $0.0036 | $3.60 | 400k tokens |
| Anthropic: Claude Sonnet 4.5 | Anthropic | $0.0040 | $4.05 | 1M tokens |
| Claude Sonnet 4.6 | Anthropic | $0.0040 | $4.05 | 1M tokens |
| Mistral Medium 3.5 | Mistral | $0.0040 | $4.05 | 256k tokens |
| Grok 4 | xAI | $0.0042 | $4.20 | 2M tokens |
| Grok 4.5 | xAI | $0.0042 | $4.20 | 500k tokens |
| Mistral Large 2 | Mistral | $0.0042 | $4.20 | 128k tokens |
| Mistral: Mixtral 8x22B Instruct | Mistral | $0.0042 | $4.20 | 66k tokens |
| Mistral: Pixtral Large 2411 | Mistral | $0.0042 | $4.20 | 131k tokens |
| Qwen 3.8 Max | Alibaba | $0.0042 | $4.20 | 1M tokens |
| Google: Gemini 2.5 Pro Preview 05-06 | $0.0045 | $4.50 | 1.0M tokens | |
| Google: Gemini 2.5 Pro Preview 06-05 | $0.0045 | $4.50 | 1.0M tokens | |
| OpenAI: GPT-5 Chat | OpenAI | $0.0045 | $4.50 | 128k tokens |
| OpenAI: GPT-5.1 | OpenAI | $0.0045 | $4.50 | 400k tokens |
| OpenAI: GPT-5.1 Chat | OpenAI | $0.0045 | $4.50 | 128k tokens |
| OpenAI: GPT-5.1-Codex | OpenAI | $0.0045 | $4.50 | 400k tokens |
| OpenAI: GPT-5.1-Codex-Max | OpenAI | $0.0045 | $4.50 | 400k tokens |
| OpenAI: GPT-3.5 Turbo 16k | OpenAI | $0.0048 | $4.80 | 16k tokens |
| OpenAI: o4 Mini Deep Research | OpenAI | $0.0048 | $4.80 | 200k tokens |
| Qwen 3.7 Max | Alibaba | $0.0052 | $5.25 | 1M tokens |
| OpenAI: GPT Audio | OpenAI | $0.0060 | $6.00 | 128k tokens |
| Gemini 3.1 Pro | $0.0060 | $6.00 | 2M tokens | |
| Google: Nano Banana Pro (Gemini 3 Pro Image Preview) | $0.0060 | $6.00 | 66k tokens | |
| OpenAI: GPT-5.3 Chat | OpenAI | $0.0063 | $6.30 | 128k tokens |
| OpenAI: GPT-5.3-Codex | OpenAI | $0.0063 | $6.30 | 400k tokens |
| Claude Opus 5 | Anthropic | $0.0067 | $6.75 | 1M tokens |
| GPT-5.4 | OpenAI | $0.0075 | $7.50 | 272k tokens |
| Anthropic: Claude 3.7 Sonnet (thinking) | Anthropic | $0.0081 | $8.10 | 200k tokens |
| Anthropic: Claude Sonnet 4 | Anthropic | $0.0081 | $8.10 | 200k tokens |
| Kimi K3 | Moonshot | $0.0081 | $8.10 | 1M tokens |
| xAI: Grok 3 | xAI | $0.0081 | $8.10 | 131k tokens |
| xAI: Grok 3 Beta | xAI | $0.0081 | $8.10 | 131k tokens |
| OpenAI: GPT-4 Turbo | OpenAI | $0.010 | $10.50 | 128k tokens |
| Anthropic: Claude Opus 4.5 | Anthropic | $0.013 | $13.50 | 200k tokens |
| Claude Fable 5 | Anthropic | $0.013 | $13.50 | 1M tokens |
| Claude Opus 4.7 | Anthropic | $0.013 | $13.50 | 1M tokens |
| Claude Opus 4.8 | Anthropic | $0.013 | $13.50 | 1M tokens |
| OpenAI: GPT-5 Image | OpenAI | $0.015 | $15.00 | 400k tokens |
| GPT-5.5 | OpenAI | $0.015 | $15.00 | 1M tokens |
| Anthropic: Claude 3.5 Sonnet | Anthropic | $0.016 | $16.20 | 200k tokens |
| OpenAI: o1 | OpenAI | $0.018 | $18.00 | 200k tokens |
| OpenAI: GPT-4 Turbo (older v1106) | OpenAI | $0.021 | $21.00 | 128k tokens |
| OpenAI: GPT-4 Turbo Preview | OpenAI | $0.021 | $21.00 | 128k tokens |
| OpenAI: o3 Deep Research | OpenAI | $0.024 | $24.00 | 200k tokens |
| OpenAI: o3 Pro | OpenAI | $0.024 | $24.00 | 200k tokens |
| GPT-5.2 | OpenAI | $0.026 | $25.80 | 200k tokens |
| Claude Mythos 5 | Anthropic | $0.027 | $27.00 | 1M tokens |
| OpenAI: GPT-5 Pro | OpenAI | $0.027 | $27.00 | 400k tokens |
| Anthropic: Claude Opus 4 | Anthropic | $0.040 | $40.50 | 200k tokens |
| Anthropic: Claude Opus 4.1 | Anthropic | $0.040 | $40.50 | 200k tokens |
| Claude Opus 4.6 | Anthropic | $0.040 | $40.50 | 1M tokens |
| OpenAI: GPT-4 | OpenAI | $0.054 | $54.00 | 8k tokens |
| OpenAI: GPT-4 (older v0314) | OpenAI | $0.054 | $54.00 | 8k tokens |
| OpenAI: o1-pro | OpenAI | $0.180 | $180 | 200k tokens |
“Needs chunking” means the model's context window can't hold this job in a single pass, so the real cost is higher than the figure shown and quality usually suffers.
Extraction is graded pass/fail on every field, so a model that is right 95% of the time still hands you 50 bad records per thousand. Price the human review time alongside the token cost — it usually dwarfs it, and a better model is the cheaper way to reduce it.
$0.0000 on Mistral: Mistral Nemo, the cheapest capable option, rising to $0.013 on Claude Fable 5 at the top end. Llama 4 Scout is the value pick at $0.0002 per run. The job is priced at 1,200 input and 300 output tokens — see the working below.
A single-page OCR'd invoice with line items ≈ 1,200 input tokens. A structured JSON record of the fields ≈ 300 output tokens.
$0.033 a month on Mistral: Mistral Nemo, $0.210 on Llama 4 Scout, and $13.50 on Claude Fable 5. Batch APIs typically halve these figures for work that can wait, and prompt caching cuts the input side further when the same context is reused.
Extraction is graded pass/fail on every field, so a model that is right 95% of the time still hands you 50 bad records per thousand. Price the human review time alongside the token cost — it usually dwarfs it, and a better model is the cheaper way to reduce it.
Yes. Every figure on this page is computed from our model pricing data, which is checked daily against each provider's official pricing page. When a provider changes a price, these numbers change with it.
Newsletter
Pricing changes, new model releases, and updated recommendations — delivered when it matters.
No spam. Useful updates only. Affiliate disclosures always clearly labeled.