LLM API pricing tracker
Input and output prices per million tokens for every model BenchLeader tracks, from first-party list prices where published and refreshed with each daily update. The ranking uses a blended price at three input tokens per output token; each model's page charts its full price history, and the last 90 days of changes are listed below the table.
As of 24 Oct 2026, the cheapest model on the pricing tracker is Gemma 4 E4B at $0.040 per million tokens (blended), followed by Llama 3.1 8B ($0.048) and Gemma 4 E2B ($0.050).
332 models · data as of 24 Oct 2026
Quality against what it costs to get there
Model size
OpenAI13
Alibaba11
DeepSeek9
Google9
Mistral AI9
Anthropic8
Meta6
Moonshot AI4
NVIDIA4
Xiaomi4
Zhipu AI4
IBM3
SpaceXAI3
Inception2
MiniMax2
Upstage2
Ant Group1- Apodex1
The dashed staircase is the efficiency frontier: at each price, the best index anyone reaches. A model above and left of the rest gives you more quality per dollar of real work. Size bands and run costs come from Artificial Analysis, which is the only source that publishes what a fixed batch of work actually cost.
- 1Gemma 4 E4B (thinking)$0.040
- 2Llama 3.1 8B$0.048
- 3Gemma 4 E2B (thinking)$0.050
- 4Granite 4.2 3B$0.053
- 5LFM2 24B A2B$0.053
- 6Qwen3.7 Flash (none)$0.055
- 7Mistral Small 3$0.058
- 8Qwen3.5-4B (thinking)$0.060
- 9Gemma 3 4B$0.063
- 10Nova Micro$0.065
- 11DeepSeek-R1-0528-Qwen3-8B$0.068
- 12NVIDIA Nemotron Nano 9B v2 (thinking)$0.070
- 13Llama 3.2 1B Instruct$0.071
- 14Sarvam 105B (high)$0.073
- 15Llama 3-8B$0.073
332 of 332
| # | ||||
|---|---|---|---|---|
| 1 | o1-pro | $262.50 | $150.00 | $600.00 |
| 2 | GPT-5.5 Proxhigh | $67.50 | $30.00 | $180.00 |
| 3 | GPT-5.4 Pro | $67.50 | $30.00 | $180.00 |
| 4 | GPT-5.2 Proxhigh | $57.75 | $21.00 | $168.00 |
| 5 | GPT-5 Pro | $41.25 | $15.00 | $120.00 |
| 6 | GPT-4 | $37.50 | $30.00 | $60.00 |
| 7 | GPT 4 0613 | $37.50 | $30.00 | $60.00 |
| 8 | o3-prohigh | $35.00 | $20.00 | $80.00 |
| 9 | Claude Opus 4.1thinking | $30.00 | $15.00 | $75.00 |
| 10 | Claude Opus 4thinking | $30.00 | $15.00 | $75.00 |
| 11 | Claude 3 Opus | $30.00 | $15.00 | $75.00 |
| 12 | o1 | $26.25 | $15.00 | $60.00 |
| 13 | Claude Fable 5.1max | $20.00 | $10.00 | $50.00 |
| 14 | GPT-6 Astramax | $20.00 | $10.00 | $50.00 |
| 15 | Claude Fable 5max | $20.00 | $10.00 | $50.00 |
| 16 | GPT-6 Astra Promax | $20.00 | $10.00 | $50.00 |
| 17 | GPT-4 Turbo | $15.00 | $10.00 | $30.00 |
| 18 | GPT-4 1106 | $15.00 | $10.00 | $30.00 |
| 19 | GPT-5.5xhigh | $11.25 | $5.00 | $30.00 |
| 20 | GPT-5.5 Instant | $11.25 | $5.00 | $30.00 |
| 21 | Claude Opus 5max | $10.00 | $5.00 | $25.00 |
| 22 | Claude Opus 4.7 | $10.00 | $5.00 | $25.00 |
| 23 | Claude Opus 4.8max | $10.00 | $5.00 | $25.00 |
| 24 | Claude Opus 4.6high | $10.00 | $5.00 | $25.00 |
| 25 | Claude Opus 4.5thinking | $10.00 | $5.00 | $25.00 |
| 26 | Claude Opus 5.5max | $8.00 | $4.00 | $20.00 |
| 27 | GPT-5.6 Solmax | $8.00 | $4.00 | $20.00 |
| 28 | Muse Wondermax | $8.00 | $4.00 | $20.00 |
| 29 | GPT-5.6 Sol Promax | $8.00 | $4.00 | $20.00 |
| 30 | GPT-5.6 | $8.00 | $4.00 | $20.00 |
| 31 | Kimi K3maxopen ↗ | $6.00 | $3.00 | $15.00 |
| 32 | Claude Sonnet 4.5high | $6.00 | $3.00 | $15.00 |
| 33 | Grok 4 | $6.00 | $3.00 | $15.00 |
| 34 | Claude Sonnet 4.6 | $6.00 | $3.00 | $15.00 |
| 35 | Claude Sonnet 4thinking | $6.00 | $3.00 | $15.00 |
| 36 | Grok 3 | $6.00 | $3.00 | $15.00 |
| 37 | Claude 3.5 Sonnet | $6.00 | $3.00 | $15.00 |
| 38 | Mistral Large 24.02 | $6.00 | $4.00 | $12.00 |
| 39 | Ember-1 | $6.00 | $3.00 | $15.00 |
| 40 |  Sonar Pro | $6.00 | $3.00 | $15.00 |
| 41 | GPT-5.4xhigh | $5.63 | $2.50 | $15.00 |
| 42 |  Nova Premier | $5.00 | $2.50 | $12.50 |
| 43 | GPT-5.3-Codexxhigh | $4.81 | $1.75 | $14.00 |
| 44 | GPT-5.2xhigh | $4.81 | $1.75 | $14.00 |
| 45 | GPT-5.2-Codexxhigh | $4.81 | $1.75 | $14.00 |
| 46 | GPT-5.3 Chat | $4.81 | $1.75 | $14.00 |
| 47 | GPT-5.6 Terramax | $4.50 | $2.00 | $12.00 |
| 48 | Gemini 3.1 Pro | $4.50 | $2.00 | $12.00 |
| 49 | Gemini Pro Latest | $4.50 | $2.00 | $12.00 |
| 50 | Gemini 3.1 Pro Customtools | $4.50 | $2.00 | $12.00 |
| 51 | GPT-4o | $4.38 | $2.50 | $10.00 |
| 52 |  Command Aopen ↗ | $4.38 | $2.50 | $10.00 |
| 53 | Command A Plus 05 2026open | $4.38 | $2.50 | $10.00 |
| 54 | Command A Reasoningthinkingopen | $4.38 | $2.50 | $10.00 |
| 55 | Gemini 4 Argonhigh | $4.00 | $2.00 | $10.00 |
| 56 | GPT-6.1 Solmax | $4.00 | $2.00 | $10.00 |
| 57 | Claude Sonnet 5.5max | $4.00 | $2.00 | $10.00 |
| 58 | NewKimi K3.1 Pro | $4.00 | $2.00 | $10.00 |
| 59 | GPT-6 Solmax | $4.00 | $2.00 | $10.00 |
| 60 | Claude Sonnet 5high | $4.00 | $2.00 | $10.00 |
Price changes in the last 90 days
Every list-price change over 1% (blended) up to 24 Oct 2026, newest first. Each model’s page charts its full price history.
| Model | Date | Before (in / out) | After (in / out) | Blended change |
|---|---|---|---|---|
| gpt-oss-20b | 24 Oct 2026 | $0.070 / $0.250 | $0.050 / $0.220 | -20% |
| GLM-4.7-FlashX | 24 Oct 2026 | $0.075 / $0.448 | $0.070 / $0.400 | -9% |
| Qwen3-Next 80B A3B | 24 Oct 2026 | $0.130 / $1.15 | $0.100 / $1.10 | -9% |
| GLM 4.7 | 24 Oct 2026 | $0.600 / $2.20 | $0.530 / $2.10 | -8% |
| Qwen3 Coder 480B | 24 Oct 2026 | $0.300 / $1.80 | $0.380 / $1.40 | -6% |
| Llama 3.3 70B | 24 Oct 2026 | $0.710 / $0.720 | $0.660 / $0.720 | -5% |
| Qwen3 235B A22B 2507 | 24 Oct 2026 | $0.220 / $0.880 | $0.210 / $0.840 | -5% |
| Qwen3.6 Max | 24 Oct 2026 | $1.32 / $7.90 | $1.30 / $7.80 | -1% |
| Qwen3.7 Flash | 24 Oct 2026 | $0.030 / $0.118 | $0.030 / $0.130 | +6% |
| Qwen2.5-Omni 7B | 24 Oct 2026 | $0.087 / $0.345 | $0.100 / $0.400 | +16% |
| DeepSeek R1 0528 | 24 Oct 2026 | $1.35 / $3.00 | $1.85 / $2.75 | +18% |
| Qwen3-VL Plus | 24 Oct 2026 | $0.143 / $1.43 | $0.200 / $1.60 | +18% |
| Qwen-Omni Turbo | 24 Oct 2026 | $0.058 / $0.230 | $0.070 / $0.270 | +19% |
| GLM 5.3 FlashX | 24 Oct 2026 | $0.299 / $1.05 | $0.370 / $1.25 | +22% |
| Qwen3.8 Flash | 24 Oct 2026 | $0.119 / $0.401 | $0.150 / $0.470 | +22% |
| Qwen3.8 Omni Flash | 24 Oct 2026 | $0.113 / $0.382 | $0.150 / $0.470 | +28% |
| GLM 5V Turbo | 24 Oct 2026 | $0.747 / $3.29 | $1.20 / $4.00 | +37% |
| Qwen3.6 Flash | 24 Oct 2026 | $0.165 / $0.990 | $0.250 / $1.50 | +52% |
| Qwen Turbo | 24 Oct 2026 | $0.044 / $0.087 | $0.050 / $0.200 | +60% |
| Qwen-MT Turbo | 24 Oct 2026 | $0.101 / $0.280 | $0.160 / $0.490 | +66% |
| Qwen3-Omni Flash | 24 Oct 2026 | $0.258 / $0.989 | $0.430 / $1.66 | +67% |
| Qwen-VL OCR | 24 Oct 2026 | $0.043 / $0.072 | $0.070 / $0.160 | +84% |
| Qwen3.5 Flash | 24 Oct 2026 | $0.029 / $0.287 | $0.100 / $0.400 | +87% |
| Qwen Flash | 24 Oct 2026 | $0.022 / $0.216 | $0.050 / $0.400 | +95% |
| Qwen3 Coder Plus | 24 Oct 2026 | $0.574 / $2.30 | $1.00 / $5.00 | +99% |
| Qwen-VL Plus | 24 Oct 2026 | $0.115 / $0.287 | $0.210 / $0.630 | +99% |
| GLM 4.5V | 24 Oct 2026 | $0.299 / $0.896 | $0.600 / $1.80 | +101% |
| Qwen3 Coder Flash | 24 Oct 2026 | $0.144 / $0.574 | $0.300 / $1.50 | +139% |
| Qwen3 14B | 24 Oct 2026 | $0.144 / $0.574 | $0.350 / $1.40 | +144% |
| Qwen3 32B | 24 Oct 2026 | $0.287 / $1.15 | $0.700 / $2.80 | +144% |
| Qwen3 235B A22B | 24 Oct 2026 | $0.287 / $1.15 | $0.700 / $2.80 | +144% |
| Qwen3 8B | 24 Oct 2026 | $0.072 / $0.287 | $0.180 / $0.700 | +147% |
| Qwen3.5 Plus | 24 Oct 2026 | $0.115 / $0.688 | $0.400 / $2.40 | +248% |
| Qwen-Plus | 24 Oct 2026 | $0.115 / $0.287 | $0.400 / $1.20 | +280% |
| QwQ Plus | 24 Oct 2026 | $0.230 / $0.574 | $0.800 / $2.40 | +280% |
| Qwen-VL | 24 Oct 2026 | $0.230 / $0.574 | $0.800 / $3.20 | +343% |
| Qwen-MT Plus | 24 Oct 2026 | $0.259 / $0.775 | $2.46 / $7.37 | +850% |
| Qwen-MT Plus | 23 Oct 2026 | $2.46 / $7.37 | $0.259 / $0.775 | -89% |
| Quasar 438B | 23 Oct 2026 | $0.600 / $1.80 | $0.110 / $0.330 | -82% |
| Mistral Nemo | 23 Oct 2026 | $0.150 / $0.150 | $0.029 / $0.030 | -81% |
And 324 earlier changes, on each model’s page.
Questions people ask
- What is the cheapest LLM API per 1M tokens?
- On 24 Oct 2026, Gemma 4 E4B by Google is the cheapest benchmarked model at $0.040 per 1M tokens blended.
- What is the cheapest good LLM?
- Among models scoring above 55 on the BenchLeader Index, Ling 3.0 Flash VL is the cheapest on 24 Oct 2026 at $0.111 per 1M tokens blended (index 55.5).
- What does blended price mean?
- One price per million tokens that weights input and output the way typical use does: (3 × input price + output price) ÷ 4, so three input tokens for every output token.
- Which LLM prices change soon?
- As of 24 Oct 2026, Gemini 4 Argon (launch price until 30 Nov 2026, then $8.00 blended), Gemini 3.8 Flash (launch price until 31 Dec 2026, then $3.00 blended) and Gemini 3.6 Flash (launch price until 31 Dec 2026, then $3.00 blended).
Cite as: BenchLeader, “LLM API pricing tracker”, data as of 24 Oct 2026, https://www.benchleader.com/leaderboards/cost.