Checked today at 12:15 UTC
The LLM deal tracker.
Every free LLM, discount, price drop and new model across 60 providers, checked every day, with the usual price next to it. In a dashboard, an API and an MCP server.
Free right now
ending soonest first- Inkling SmallOpenRouter, served by Thinking MachinesNov 4ends tomorrow
- InklingOpenRouter, served by Thinking MachinesNov 4ends tomorrow
- Dots3-Note PreviewOpenRouter, served by AtlasCloudDec 31
- FlintOpenRouter, served by Recursiveno end date
- Xing4.0 29B A4BOpenRouter, served by Phalano end date
All 19 free models ›free endpoints carry rate limits
On sale today
The biggest discounts right now
| Model | Deal | Now, input · output | List price |
|---|---|---|---|
| GLM-5.3Z.ai | −80%at Baidu | $0.28 · $0.88 | |
| Ling 3.1 FlashinclusionAI | −75%at DeepInfra | $0.075 · $0.23 | |
| Qwen3 235B A22B Instruct 2507Alibaba | −75%at GMICloud | $0.0875 · $0.35 | |
| DeepSeek V4 Flash 0731DeepSeek | −70%at StreamLake | $0.13 · $0.40 | |
| GLM 5.2Z.ai | −56%at StreamLake | $0.61 · $1.93 |
Price paid today against the same host's list price, input + output per million tokens · the list price is what you pay when the discount ends.
New releases
Latest LLMs, with their real prices
Also recently: Mercury 3 · DeepSeek V4.1 Flash 1017 · Muse Spark Open · Nemotron 3.5 Super VL · Muse Wonder
Price drops
Biggest price drops this week
| Model | Provider | Input · output, before | Input · output, now | Change ↓ |
|---|---|---|---|---|
| GLM 5.3 | Baidu | $0.28 · $0.88 | −80% | |
| Nemotron 3.5 Super VL | Venice | $0.20 · $0.50 | −53% | |
| DeepSeek V4.1 Flash 1017 | StreamLake | $0.15 · $0.60 | −50% | |
| Mistral Large 4 | Mistral AI | $0.70 · $2.10 | −49% | |
| GLM 5.3 | Reka | $0.14 · $2.10 | −48% |
From 5% on the input + output price · offers that reprice often and changes of source are left out · per million tokens.
Intelligence for the price
The cheapest way to each level today
| Level (Epoch) | Cheapest model | Provider | Score | Input + output |
|---|---|---|---|---|
| Tied with the best | GPT-6.1 Sol | OpenAI | 168 | $12.00 |
| 160 and up | Claude Haiku 5.5 | Anthropic | 160 | $0.60 |
| 155 and up | DeepSeek V4.1 Flash | Morph | 156 | $0.35 |
| 150 and up | GLM-5.3-Flash | Pareto Inference | 153 | $0.26 |
| 145 and up | Same model as the level above: nothing cheaper reaches this level. | |||
Score: Epoch Capabilities Index by Epoch AI (CC BY 4.0) · « Tied with the best »: the model's interval reaches the leader's · price: cheapest offer, input + output per million tokens, discounts included.
For developers
One question, three ways to ask it
Where is GLM-5.3 cheapest today?
Dashboard
Models › GLM-5.3 › offers
- Baidu$0.28 · $0.88
- Morph$0.087 · $1.40
- Reka$0.14 · $2.10
API
One open JSON file, refreshed every day
curl -s https://apipriceindex.com/api/all.json \
| jq '[.models[]
| select(.base_model=="glm-5.3" and .pricing)]
| min_by(.pricing.input_per_mtok
+ .pricing.output_per_mtok)
| {name, input: .pricing.input_per_mtok,
output: .pricing.output_per_mtok}'
{ "name": "GLM 5.3 (Baidu)",
"input": 0.28, "output": 0.88 }MCP server
Your agent asks, the index answers
pip install apipriceindex-mcp
claude mcp add apipriceindex -- apipriceindex-mcp
> Where is GLM-5.3 cheapest?
31 hosts. Cheapest: Baidu,
$0.28 in, $0.88 out per M, read Nov 3.Why trust it
Numbers you can check
Every price has a source
Read on the provider's own price list or API, with the date we read it. Anything odd waits for a human check.
The usual price next to it
A discount is shown against the host's list price, a drop against the daily history of the same offer. Offers that reprice all the time are flagged, not counted.
Prices from the provider
848 offers read every morning from each provider's catalog or pricing page, and checked against OpenRouter, models.dev and LiteLLM.
Open data
JSON, CSV and the MCP server under CC BY 4.0. Use it, cite it: use this data.
Models ›
210 models, one page per variant, with every provider that serves them.
Providers ›
60 providers: what they serve, at what price, under which terms.
Guides ›
Guides, analyses and rankings, computed from the index.